The Key Requirements for the Greening of Emerging IT Technologies: A Comprehensive Systematic Review

1. Introduction

The human ecosystem has had two significant implications, as an outcome of worldwide industrial growth; the first is the depletion of natural resources, wherein the consumption is outpacing its renewal (Gielen et al., 2019; Akadiri et al., 2022). Stepping-up operations and searching for alternate energy measures and sources are prima; for retarding the current rate of resource depletion (Owusu & Asumadu-Sarkodie, 2016). Secondly, global industrial operations have led to manifold carbon emissions (Ahmed et al., 2022), known as Greenhouse Gases (GHG); which further lead to higher diseases, global warming and depletion of the protective Ozone layer (Gavurova et al., 2022).

The Information Technology (IT) industry besides being an emerging global IT is both an emerging global industrial vertical and has the potential to support distinct business verticals (Ali et al., 2020). The pace at which we demand information and the granularity of it is momentous, and how it eases our life are also colossal. However, it is to be noted that the IT industry, especially the general computing industry in general, contributes immensely to energy consumption and thus carbon emissions (Shuja et al., 2017; Belkhir & Elmeligi, 2018). Thus, there lies a need for intervention from multiple nodes like IT technologies, work processes, algorithms, infrastructure design and utilization and others; by which each of them needs to be redefined for energy efficiency, commensurate consumption and designing sustainable work processes (Koronen et al., 2020). Moreover, IT industries need to exhibit and enable responsible consumption; by just not reducing their carbon footprint, but also structuring systems such that industries from another sector can also foster green practices, in their routine operations (Freitag et al., 2021; Shuja et al., 2016). The advent of the IT industry has revolutionized the planet, communities and societies that we dwell in. The challenge of sustainability and energy consumption and aggravated deficit is attributed to the aviation, petroleum and manufacturing sectors (Barke et al., 2022); the major accountability also lies with the IT sector, owing to their escalated consumption patterns and carbon emissions (Freitag et al., 2021). Globally, the IT sector alone guzzles 2.4% of the consumption, which is forecasted to increase to 20%, annually (Koot & Wijnhoven, 2021). Concurrently the sector also equally contributes to 2.5% of the global carbon emissions, which equates to around 0.86 metric gig tonnes of CO₂ (Freitag et al., 2021; Bressler, 2021). This exorbitant and fatal consumption pattern and embedded carbon emissions necessitate ‘Green’ computing infrastructure and processes (Lannelongue et al., 2021).

Green computing is conceptualized for maximizing efficiency, exploring new energy sources, responsible consumption, and ‘re-usability’, wherever possible. It calls for the use of sustainable materials/products for manufacturing practices; and the adoption of green initiatives in an array of industries, which are supported by IT management and monitoring tools (Nwankwo & Chinedu, 2020). The duality of resource efficiency for performance and energy consumption is pivotal. High-performance computing aggregates the use of diverse sources and alternative backup patterns (Roelich et al., 2015); which is exactly in contravention to the principle of energy efficiency. These practices necessitate the reduction of utilized resources and energy-proportionate computing methods. In the computing milieu, energy reuse can be emancipated as a cyber-physical association of IT resources, besides cooling of large-scale IT centres (Shuja et al., 2016; Bharany et al., 2022). Primarily, the use of renewable energy sources for computing and diminishing carbon footprint is required (Freitag et al., 2021). Industrialized operations that source e-waste and recycle/reuse the components can also significantly ameliorate the concern (Abdulaziz, 2022). Besides these, the allied industries are emulating the green-computing practices of IT industries; besides proliferating these initiatives in other industries/sectors, through systematic environmental monitoring practices and consciousness (Deng et al., 2014).

The recent past has witnessed the emergence of multiple technologies that enable scientific, industrial and social businesses through sustainable IT practices. Their induction into human lifestyles has been an outcome of their seamless offerings, about ease of communication, health monitoring and environmental safeguards (Peng et al., 2020; Benis et al., 2021). The current manuscript coagulates the published literature, around the concept of Green Computing, from the lens of emerging IT and green technologies. The current review handpicks cloud computing (Ali et al., 2020), mobile computing (Al-Ahmed et al., 2021), Internet of Things (IoT) (Zhang et al., 2021), big data analytics (Calza et al., 2020), software-based networks (Qureshi et al., 2020), blockchain technology (Diniz et al., 2021), and artificial intelligence (AI) (Dauvergne, 2022) as the benchmark IT technologies for review, here. The selection is based on metrics set for popularity, social integration, and future applications in smart environments (Keeler & Bernstein, 2021).

This review differentiates itself in encapsulating distinct IT Technologies, which could act as a reference to organizations in creating opportunities and understanding these technologies, to deliver their responsibilities well. So, we look forward to a three-fold contribution: First, aggregate decadal information on green technologies, their circuits, algorithms and architectures; and emerging IT technologies; second, underscore the infrastructure and inputs for the greening of these technologies; and lastly, accentuate trending research themes on the greening of this computing – IT Technologies.

The manuscript furnishes a prologue to the greening of trending IT technologies; by way of detailed review, planning and sustained methodology. Additionally, the structured classification framework that themes the detailed review can aid future cult of researchers in mapping understanding across the existent literature; and to identify gaps for future research.

2. Emerging IT Technologies Background

Cloud computing, mobile computing, IoTs, big data, blockchain, software-based Networks, and AI are popular and widely used technologies and are often interdependent. They are expected to provide people and organizations with more reliable and intelligent services, and sustainable answers for several organizations' services and administrative challenges. This segment provides a succinct synopsis of these emerging IT trends.

2.1. Cloud Computing

Cloud computing is the recent breakthrough in Internet-based computing technologies that offer on-demand infrastructure, software and platform services on a pay-as-you-go basis (Jamal & Khan, 2020; Banger et al., 2022). In this era of computing technology, the cloud presents a flexible execution environment of assets from multiple partners and provides metered administration at various granularities (Grace & Gandhi, 2022). Jamal and Khan (2020) demarcated cloud computing as a modelling mechanism that offers ease of access and convenience to an authorized pool of configurable computing infrastructure of integrated server applications and embedded IT assistance that can be dispensed with negligible management intervention. Grace and Gandhi (2022) referred to cloud computing as special expert collaboration, where massive virtualized and flexible work is utilized through the internet. While Ogala et al. (2022) presented cloud computing as a set of network-based services with ubiquitous and on-demand network access to a set of programmable computing resources with scalable, cost-effective and low management efforts. The National Institute of Standards and Technology (NIST) identified five main characteristics of cloud computing such characteristic as offering broad network access, on-demand self-service, real-time scalability, resources pooling, and measured service throughout the utilization of pay-per-use concept (Ali et al., 2021). Other characteristics, such as flexibility, reliability, universality, inexpensive, and unlimited storage are identified by different researchers (Jamal & Khan, 2020; Grace & Gandhi, 2022; Banger et al., 2022)

Nowadays, cloud computing has become one of the dominant computing technologies to achieve substantial scale (Jamal & Khan, 2020). It has been utilized by both individuals and business organizations to run applications store data over the cloud and access them from anywhere and at any time of internet connection (Banger et al., 2022). Accordingly, the influence of cloud computing on businesses and individuals is obvious. For instance, cloud computing offers several benefits for individuals, such as huge amounts of online storage, high availability of everything at the touch of a button, high-end machines, and additional computational power in sort of virtual machines (Banger et al., 2022). Freelancer software developers now have the privilege to create applications and internet services to be offered worldwide (Banger et al., 2022). Researchers now can share and analyse data at a large scale (Banger et al., 2022). Moreover, the presence of the operated software over the cloud has changed many elements in business organizations’ start-up and daily activities (Banger et al., 2022). Organizations can cut off their costs and expand their offering by using cloud services instead of purchasing and maintaining expensive hardware and software. Furthermore, cloud computing can significantly support the green environment due to the sharing of resources, which allows the consumption of smaller amounts of power and energy, and hence, carbon emission is reduced accordingly (Jamal & Khan, 2020).

2.2. Mobile Computing

Mobile computing is another emerging technological trend in the computing paradigm, where its applications are involved in multiple disciplines and domains (Elazhary, 2019; Nazir et al., 2019). The term “mobile” typically represents smartphones, yet it applies to all portable, wireless, programmable, and handheld devices that can be used anywhere and at any time (Zaidi et al., 2021). These devices include smartphones, tablets, smartwatches, and laptops. Smartphone devices have become an essential part of people's daily life activities. They are very useful for accomplishing several on-the-go activities of social communication and interactivity, data storage, and online shopping (Elazhary, 2019).

Mobile computing provides remote and quality information services such as processing, storage, information and queries to mobile users (Nazir et al., 2019). It is a type of technology that enables computers and information devices to share data without the need for a physical connection (Ma et al., 2018). It is also can improve the accuracy of identifying relevant information and increase the productivity’s efficiency (Ma et al., 2018). Accordingly, various mobile applications have been and are being developed by the research community for various application domains such as context-aware mobile applications, mobile health (m-health), mobile learning (m-learning), mobile government (m-government) (Nazir et al., 2019; Zaidi et al., 2021).

2.3. Internet of Things (IoT)

The IoTs is yet another breakthrough technology that is inviting the attention of the IT research fraternity; and has the potential to be one of the dominant research paradigms soon (Elazhary, 2019; Ramasamy & Kadry, 2021). IoT as a term was coined by Kevin Ashton in 1999, due to the linkage between Radio Frequency Identification systems (RFIDs) and sensors to the internet (Elazhary, 2019). In this linkage, data about things in the real world is gathered directly by computers to exceed the human-provided data limitation.

Due to this smart privilege of accessibility of smart devices with low cost, many researchers envisioned a smart world embedded with sensors, actuators, or other wearable devices connected to the internet to shape IoT applications, which aim to automate the extraction of data internet-connected things and to control them in an informed manner (Elazhary, 2019; Liao et al., 2020). IoT represents an evolutionary stage of the internet, which makes a global communication infrastructure between humans and machines more effective and automated (Ramasamy & Kadry, 2021). It constructs the global infrastructure that will change the fundamental aspects of our lives, starting from health services to manufacturing and supply chain, and from agriculture to mining (Ramasamy & Kadry, 2021). IoT appeared as a powerful technique with appliances in various domains like smart grid, smart city, smart industry, smart building, smart health, smart energy, and smart environment, water and food monitoring (Liao et al., 2020). IoT offers many technical benefits, including the ability to locate and trace in a short amount of time, intelligence swap and management of information, enhanced power solutions and decreased waste (Ramasamy & Kadry, 2021). Moreover, IoT comes with various characteristics such as safety, scalability, dynamicity, interconnectivity, and service-related things (Ramasamy & Kadry, 2021). Accordingly, mobile and cloud computing are inseparable technologies from the IoT due to the various services, applications, and devices offered to the IoT technology (Ma et al., 2018; Nazir et al., 2019). For instance, IoT through its smart sensing technology provides vital support for mobile medical data transmission and acquisition in real-time environmental conditions (Ma et al., 2018).

2.4. Big Data Analytics

Big data is a term that is used to refer to a collection of large, growing, and complex data sets that originated from heterogeneous and autonomous sources and require powerful technologies and advanced algorithms to process them (Awan et al., 2021; Wang et al., 2021). These datasets are very big as they are measured in Exabyte (Giri & Lone, 2014). Unlike traditional data, big data includes masses of structured, unstructured and semi-structured data that can be in the form of text, pictures, videos, maps and so on (Oussous et al., 2018). A few examples of big data include sensor networks, environmental sensors, patients' electronic records, bank records, online posted videos and photos, social media comments, and customer reviews and comments on e-commerce (Giri & Lone, 2014). The majority of data experts identify big data by six main characteristics volume, velocity, variety, veracity, valence, and value (Awan et al., 2021; Oussous et al., 2018). Due to this complexity, the traditional and existing database management and analytical tools became insufficient to handle and process big data (Haoxiang & Smys, 2021; Ageed et al., 2021). However, once advanced data management and analytics technologies are adopted, organizations can have new pivotal business opportunities through the development of innovative products and services, and the achievement of more effective and efficient business operations and decision-making (Oussous et al., 2018).

Academic and business communities identified big data as a breakthrough technological paradigm that dramatically can inspire public and private organizations in several application domains such as e-commerce, e-government, science, health, security, transportation and logistics, and IoT (Ageed et al., 2021; Haoxiang & Smys, 2021; Awan et al., 2021). In today’s technology and knowledge-driven society, the intersection of big data and data mining is a strategic tool and a key development factor toward organizations’ success, upgrade, and growth (Ageed et al., 2021). Due to this intersection of advanced analysis with training-calculating technology over a large pool of resources, the most valuable information and expertise are extracted and delivered to the user as a service for better decision-making in business intelligence and other applications of network instruction, spam filtering, recommendation systems, and health analytics (Awan et al., 2021).

2.5. Blockchain Technology

Blockchain is one of the emerging computing technologies that aims to store and transmit data in a more secure, transparent and decentralized way (Ali et al., 2021; Riđić et al., 2022). It has been widely accepted as a public or private shared ledger of all digital events or as a distributed database shared or run among the participating agents in the blockchain (Casino et al., 2019; Jaoude & Saade, 2019). From the technical aspect, blockchain is a distributed system in the form of blocks, which contains all the transactions done since its start and implements a set of agreed rules where neither the user nor the system operators can break them no one (Issaoui et al., 2019). Blockchain technology is associated with decentralization, anonymity, proof tampering, and audibility that offers a secure data management and transparent modelling process in such areas as supply chain management, emission trading, health care, smart city, and food and energy traceability (Issaoui et al., 2019; Ali et al., 2021).

Blockchain is one of the most recent technologies developed that emphasizes the innovations of the IoT, e-commerce, telecommunication, and AI revolutions (Riđić et al., 2022; Jaoude & Saade, 2019). It is also one of the recently developed technologies that may create new opportunities for all sectors, such as finance, supply chain, healthcare, government, education, accounting & auditing, e-voting, asset management, and identity management (Riđić et al., 2022; Jaoude & Saade, 2019). Such opportunities can be in sort of enhancing transparency, enhancing and fastening business processes, building trust, reducing cost, and ensuring security (McGhin et al. 2019; Katuwal et al., 2018). For instance, blockchain could play a vital role in records management and data sharing in the healthcare sector, or by allowing users to validate and synchronize the contents of a transaction ledger in the financial sector (Jaoude & Saade, 2019).

Several research studies have highlighted the potential benefits that may be observed by varied sectors and industries through the utilization of blockchain technology (Ali et al., 2021). For example, a study conducted by Beck et al. (2018) classified blockchain as a distributed ledger technology that provides confidence to the users. Moreover, various studies have shown that blockchain comes with the capability to decrease transactional obscurity, insecure states, and dubiousness, and to renovate economies and the social order profoundly through a reduction in the cost of transactions and the requirement for well-recognized and trustworthy third parties (Nærland et al., 2017; Clemons et al., 2017). Despite the several benefits offered by blockchain, some sectors as government and education still might see blockchain as a distracting technology that cannot be trusted and needs to be managed (Ali et al., 2021). Indeed, blockchain is still a new technology, and while there is considerable excitement regarding its potential benefits, there is also considerable inaccurate information and uncertainty regarding the potential utility of blockchain in general (Ali et al., 2021). For example, the application of blockchain in the circular economy (CE) is a promising field of study with enormous potential for achieving sustainability and green computing development (Francisco Luis et al., 2022). Blockchain technology comes with decentralization, anonymity, proof tampering and audibility features, which can be used in related cases to CE principles such as food and energy traceability, agriculture processes, and supply chain management. Blockchain can enable any CE project to obtain a secure platform of data management with transparent processes for end-users (Francisco Luis et al., 2022).

2.6. Artificial Intelligence (AI)

AI is the science of creating smart devices and machines in the form of different computer programs (Ali et al., 2023). It is also known as the technology of making computers so intelligent in a way that they can understand humans (Androutsopoulou et al., 2019). AI is one of the major turning points in the computer science field and revolutionary industry that allows developers to create intelligent machines that behave like humans, think like humans and make their own decisions (Vesnic-Alujevic et al., 2020; Dwivedi et al., 2021). From the general perception, AI is defined as the study, design and development of intelligent agents of computational systems that recognize their environment to be capable of taking actions and decisions to maximize the chances of success (Vesnic-Alujevic et al., 2020; Zuiderwijk et al., 2021). With the ultimate goal of creating consciousness, AI goes through reasoning, learning, perception, prediction, planning or controlling phases (Zuiderwijk et al., 2021).

Recently, AI has been recognized as a revolutionary computational system field for society's transformation toward automation and intelligence, which offers competitive solutions to several services and administrative challenges, regardless of the adopted industry or sector (Dwivedi et al., 2021; Alshahrani et al., 2021). According to Alshahrani et al., (2021), the global spending on AI by 2023 is expected to be around 98 billion dollars, which is approximately double the amount spent by 2019. This transformational impact of AI has led to significant academic and industrial interest. In recent years, AI techniques have been widely used in private and public sector services to support automation and to enhance the quality of decision-making as well as problem-solving through the implementations of such techniques and tools as natural language processing, conversational agents, robotics, expert systems, neural network, fuzzy logic, and machine learning (Alshahrani et al., 2022; Androutsopoulou et al., 2019)

AI applications offer varied benefits through their utilization in several public and private sectors such as healthcare, banking, media, tourism, retail, stock market, telecommunication, education, security, social welfare, public safety, supply chain and communications (Sousa et al. 2019; Androutsopoulou et al., 2019). Among the benefits that are offered by AI applications to these sectors are reducing cost, increasing productivity, decreasing dependency on human decision-making, solving resource allocation problems, creating new employment opportunities, improving public service delivery, improving citizen’s and customer satisfaction, improving employee’ workload, reducing the administrative burden of public organizations, advancing the communication channels between government and citizen or between private organizations and customers (Androutsopoulou et al., 2019)

3. Review Planning and Methodology

According to Kitchenham et al., a systematic review (SR) involves systematically identifying, evaluating, and interpreting accessible and available literature pertaining to a specific research question, topic, or area of interest (Kitchenham & Charters, 2007). It serves as a methodical approach to gathering, organizing, and evaluating existing research, ultimately pinpointing gaps in the current literature and suggesting potential avenues for future research (Dabić et al., 2020; Paul et al., 2021). By comprehensively exploring research directions, identifying shortcomings, and proposing potential research themes, SR contributes to advancing knowledge in various domains (Khatoon & Rehman, 2021).

Differentiated into domain, theory, and methodological reviews and various subcategories, SR plays a crucial role in addressing tailored challenges and is widely adopted across industrial sectors, particularly in energy and IT industries (Kamboj & Rahman, 2015). It serves as a valuable reference for understanding sector trends, shaping policies, and guiding professional practices (Moher et al., 2009). This systematic review is grounded in structured recommendations outlined by Watson (2015), which detail the comprehensive process of searching, executing, and reporting for such reviews. For professionals in IT and energy sectors, navigating through a vast array of volatile publications to inform evidence-based decisions can be daunting (Bastian et al., 2010). However, leveraging SR can provide a holistic, unbiased, and informed view of research studies to facilitate more robust decision-making and serve as a cornerstone for evidence-based practices (Abbas et al., 2008).

In order to ensure the rigor and effectiveness of a systematic review, it is essential to follow established protocols and guidelines that prioritize reproducibility, objectivity, transparency, and rigor (Boell & Cecez-Kecmanovic, 2015). By adhering to structured protocols proposed by leading scholars such as Kitchenham, Charters, Ali, and Watson, the SR process can be elevated to a standard of excellence (Tranfield et al., 2003). It is imperative to define clear objectives, articulate research questions, select appropriate methodologies, and establish a robust classification framework at various stages of the review process (Kitchenham & Charters, 2007; Ali et al., 2020). Implementing strict search criteria, employing quality assessments, and summarizing discussions are essential steps that should be meticulously executed to ensure the quality and integrity of the SR (Watson, 2015). The specific guidelines and procedures outlined in Figure 1 for this SR aim to provide a detailed roadmap for conducting a thorough, high-quality systematic review.

3.1. Planning Stage

At the outset, this initial phase of the systematic review (SR) revolves around recognizing the necessity for conducting the review, with a focus on objectively synthesizing existing information related to a particular subject or phenomenon. Despite the presence of prior research, a lack of comprehensive systematic literature reviews persists due to various contributing factors. Conducting an SR serves the purpose of organizing these findings and literature, thereby facilitating a comprehensive analysis of existing theoretical and practical knowledge.

Following the establishment of the review's need, the next crucial step involves formulating precise research questions, which are fundamental to any SR (Paul et al., 2021). For the present SR, two central questions have been crafted: What are the essential prerequisites for the integration of environmentally friendly practices in emerging IT technologies? What are the findings of previous studies in this area, and what implications do they have for future research endeavors?

Subsequently, in the planning phase, the selection of relevant articles is executed as the third sequential step, where the article selection strategies aim to provide direct literature and evidence pertinent to the research questions. To mitigate biases, the initial selection of articles is based on defined protocols, with room for refinement during the search process (Dabić et al., 2020). Employing an integrated search strategy facilitated an automated search across various online repositories, complemented by manual screening of the retrieved articles.

An elaborate automated search strategy is deployed to encompass the most pertinent online repositories (Golder et al., 2014; Rosado-Serrano et al., 2018). For this systematic review, databases such as Direct, ACM Digital Library, IEEE Xplore, Emerald Insight, and Scopus were selected as relevant sources. A diverse array of filtering mechanisms and tools were applied to obtain relevant outcomes (McLean & Antony, 2014), followed by manual assessment. The iterative process involved initial screening based on article titles and subsequent abstract readings (Golder et al., 2014). This was followed by a thorough examination of the complete content of shortlisted articles to filter out irrelevant material (Ali et al., 2018). The utilization of the backward snowball technique in the third iteration entailed scrutinizing the references of reviewed articles to identify additional pertinent literature or seminal works (Wohlin, 2014), ensuring alignment with key research criteria such as language, temporal relevance, and publication nature. Articles that did not meet these criteria were excluded, with previously reviewed and redundant articles also eliminated. The remaining articles were considered for potential inclusion in the review study.

Table 1. Selection criteria.

Criteria	Inclusion	Exclusion	Rationale
Type of publication	Scholarly articles	Reports and any other sources	To ensure that the research retrieves information of academic level sources.
Peer-reviewed	Peer-reviewed	Non-peer-reviewed	To ensure the high quality of the used articles.
Publication year	Articles published from 2012 to 2022	Articles that published prior to 2012	To ensure the validity of the content in any article used in this research review. The pace of technology changes is relatively rapid and the past 10 years is an appropriate time period when the authors can observe the recent trends.
Language	English language	Any language other than English	English is the official language of research articles.

Table 1. Selection criteria.

Criteria	Inclusion	Exclusion	Rationale
Type of publication	Scholarly articles	Reports and any other sources	To ensure that the research retrieves information of academic level sources.
Peer-reviewed	Peer-reviewed	Non-peer-reviewed	To ensure the high quality of the used articles.
Publication year	Articles published from 2012 to 2022	Articles that published prior to 2012	To ensure the validity of the content in any article used in this research review. The pace of technology changes is relatively rapid and the past 10 years is an appropriate time period when the authors can observe the recent trends.
Language	English language	Any language other than English	English is the official language of research articles.

The subsequent phase, namely the fourth stage, involved the development of a research review protocol, which serves as a fundamental component for gaining insight into the current landscape of theoretical and practical viewpoints. A systematic review (SR) necessitates a predefined methodology to minimize the potential for bias. The present review study adopts a modified version of the comparative classification framework utilized in a previous IT literature review by Ali et al. (2018). Within this review, a tailored categorization framework encompasses seven specific technologies, each with distinct categories, and each category outlines various essential environmental considerations. The technologies and their respective categories are outlined as follows: Cloud computing technology: Encompassing cloud data centers (Ali et al., 2020). Mobile computing: Encompassing smartphones and mobile cloud computing (Al-Ahmed et al., 2021). Internet-of-Things (IoT): Encompassing Radio Frequency Identification (RFID), Wireless Sensor Networks (WSN), and Machine to Machine Communication (M2MC) (Zhang et al., 2021). Big data analytics: Encompassing data management systems (Calza et al., 2020). Networking: Encompassing network systems (Qureshi et al., 2020). Blockchain technology: Encompassing various use cases (Diniz et al., 2021). Artificial Intelligence (AI): Encompassing neural networks (Dauvergne, 2022).

All selected articles included in our review addressed these seven technologies outlined in the classification framework, namely cloud computing, mobile computing, Internet-of-Things, big data analytics, networking, blockchain technology, and AI. For detailed information on the steps and processes involved in article selection, please refer to the subsequent sections.

3.2. Execution Stage

The execution phase of our review study involved implementing the methodologies defined in the planning phase to identify relevant articles. The specific steps undertaken in this stage are as follows: Identification of Search Terms: An ongoing process of identifying search terms was conducted by compiling unique terms from seminal articles in the field (Hu & Bai, 2014; Paul et al., 2021). This process continued until all prominent terms were exhausted, adhering to the principles outlined earlier. The selected databases featured advanced search capabilities, allowing the use of both single and combined search terms. The keywords identified included: “information system” OR “information technology” OR “emerging technologies” OR “cloud computing” OR “mobile computing” OR “Internet of Things” OR “mobile technologies” OR “big data analytics” OR “soft-based networks” OR “blockchain technology” OR “artificial intelligence” AND “green technology” AND “requirement” AND “power” OR “energy”. Application of Filtering Tools: Filtering tools were applied to the databases to optimize search results (Zhang et al., 2014). Filters included thematic restriction (IT & energy), temporal restriction (publications between 2012-2022), document type restriction (published journal articles/conference papers), and linguistic restriction (English language publications). Manual Relevance Check: A manual review of titles and abstracts was performed to ensure relevance (Pucher, 2013; Ali et al., 2018). Relevance Analysis: Articles were further analyzed for their relevance to the research topic and theme (Ali et al., 2021). Backward Snowball Technique: This technique was employed to uncover articles that the automated search might have missed (Spanos & Angelis, 2016). Quality Assessment: A quality assessment criterion, adapted from Hu & Bai (2014), Ali et al. (2018), and Sadoughi et al. (2020), was applied. This included criteria such as: clarity of the research problem and questions, availability and description of data, thorough elaboration of methodology, and comprehensive presentation of research results.

This review utilized quality scores to evaluate the relevance of the results, ensuring that individual quality factors, such as validation methods, aligned with the study's objectives. After an initial selection, the quality assessment aimed to eliminate bias and enhance the validity of the systematic review. A total of 374 articles were identified and assessed based on criteria such as diligence, reliability, accuracy, and propriety to ensure relevance to research concepts and terms. Further criteria like originality, target relevance, and usability for future researchers and practitioners were also considered. The selected articles were taxonomically categorized based on primary research aims, methods, contributions, and results, facilitating the identification, extraction, classification, and synthesis of relevant data pertaining to research issues.

The review was conducted from January 11th, 2022, to May 10th, 2022, in accordance with the protocol established during the planning stage. The initial search using the defined keywords yielded a total of 1988 articles. Following the application of all the outlined steps, 374 research articles met the quality assessment criteria.

3.3. Summarizing Stage

The final selection of articles for this review study is summarized in Table 2. Initially, the keyword search identified 1988 unique articles. Applying the specified criteria reduced this number to 1250. A subsequent manual review focused on filtering out irrelevant articles, specifically targeting empirical and conceptual papers, resulting in the elimination of 568 articles and leaving 682 articles for further review.

An exhaustive review process followed, involving a thorough examination of each article’s objectives, research questions, research methods, and the quality of data and analysis techniques. This detailed reading identified 229 irrelevant articles, reducing the total to 453. The application of the backward snowball technique added 31 additional articles, bringing the total to 484. Finally, after applying the quality assessment criteria outlined in the previous stage, 110 articles were eliminated, resulting in a final count of 374 articles included in the review.

To refine the extensive pool of 1988 academic articles down to the final 374, this review study employed multi-step selection process emphasizing both relevance and quality. This review approach is outlined as follows: (1) Initial screening of the title and abstract, this step conducted a preliminary review of the titles and abstracts of all 1988 articles. Articles that clearly did not pertain to the greening of emerging IT technologies were excluded. This included articles that focused solely on non-IT-related environmental studies or those unrelated to sustainable practices in IT. (2) Inclusion and exclusion criteria, and this include inclusion criteria; articles were considered if they addressed at least one of the seven key IT technologies (cloud computing, mobile computing, Internet of Things, big data analytics, networking, blockchain technology, and AI). Discussed aspects of energy efficiency, reduction of carbon emissions, or sustainable IT practices. Presented empirical research, case studies, or comprehensive reviews relevant to green IT technologies. Exclusion criteria: articles were excluded if they lacked a clear focus on environmental sustainability within the context of IT. Articles were purely theoretical without empirical data or practical applications. Also, articles were duplicates or preliminary versions of included studies. (3) Quality assessment, and this include peer review status, in this step reference was given to peer-reviewed journal articles, conference papers, and high-impact academic publications. Research methodology, in this step the review assessed the robustness of the research methodologies, prioritizing studies with clear, replicable methods and those that included quantitative data, experiments, or significant case study insights. (4) Relevance to greening technologies. In this step the review include specificity to key technologies, articles were further scrutinized to ensure they provided detailed insights or advancements specific to the seven key technologies identified. Alignment with review objectives, the review ensured that the selected articles aligned closely with the objectives of our review, focusing on energy consumption sources, greenhouse gas emissions, and potential energy-saving techniques across different IT technologies. This systematic and thorough filtering process ensured that the final 374 articles were highly relevant, of significant academic quality, and provided comprehensive coverage of the key requirements for the greening of emerging IT technologies.

3.4. Common Characteristics related to Selected Articles

3.4.1. Temporal distribution

The earliest literature on the topic of greening emerging IT technologies was published in 2012 (see Figure 2). The peak publication frequency occurred in 2018 with 63 articles, while the lowest frequency was in 2012 with 15 articles. The majority of publications were concentrated between 2017 and 2021, reflecting a recent interest in this research area.

3.4.2. Distribution of Articles by Database

Figure 3 illustrates the databases selected for this review study, including Science Direct, the ACM Digital Library, IEEE Xplore, Emerald Insight, and Scopus. The highest number of selected articles, 94, were sourced from the Scopus database, followed by 88 articles from Emerald Insight. Additionally, 76 articles were obtained from IEEE Xplore and 68 from Science Direct. The ACM Digital Library contributed the fewest articles, with a total of 48.

3.4.3. Article distribution Affording to Classification Framework

The review is systematically categorized into seven distinct technologies: cloud computing, mobile computing, the Internet of Things, big data analytics, networking, blockchain technology, and artificial intelligence. Figure 4 presents the annual publication count for articles related to each technology. Overall, the number of articles published is as follows: cloud computing (n=75), mobile computing (n=77), Internet of Things (n=74), big data analytics (n=57), networking (n=45), blockchain technology (n=30), and artificial intelligence (n=30).

3.5. Research Synthesis and Propositions

In this SR, we have identified the key requirements for greening the emerging IT technologies. A multitude of different theoretical approaches are stated to classify and categorize the emerging IT technologies, on the basis of prior scientific research and theories (Brem & Viardot, 2015; Hamdan & Yahaya, 2016). This review study reveals seven different technologies, these technologies and their categories and the key requirements are some of most important hierarchical and interconnected for many different organizations within different sectors.

The research framework considered seven different technologies, namely cloud computing, mobile computing, Internet-of-Things, big data analytics, networking, blockchain technology, and AI. In addition, this review study identified several different categories with their key requirements to be greening technologies. All these technologies and their categories as well as their key requirements were grouped (see Table 3). The seven main identified technologies and their categories were classified based on their (1) Popularity which refers to the extent to which each technology is being adopted and implemented in various sectors. (2) Social integration which refers to the degree to which these technologies are embedded into daily life and organizational processes. (3) Future applications in smart environments which refers to the potential for future advancements and applications in creating smart, sustainable environments. These criteria were selected to ensure that the technologies included are relevant, widely adopted, and have significant potential for future development and impact. Examples of greening each technology are as follows: Cloud computing; utilization of energy-efficient data centers, adoption of renewable energy sources, and implementation of advanced cooling techniques. Example for this is Google’s data centers use advanced cooling systems and artificial intelligence to optimize energy usage, reducing their carbon footprint significantly. Mobile computing; development of energy-efficient mobile devices, use of sustainable materials, and improved battery technologies. Example for this is Apple’s iPhone recycling program and use of recycled aluminum in device manufacturing reduce e-waste and promote sustainability. IoT; deployment of low-power sensors, energy-harvesting devices, and smart grid technologies. Example for this is smart thermostats like Nest reduce energy consumption by learning user preferences and optimizing heating and cooling. Big data analytics: optimization of data processing algorithms to minimize energy usage and employing green data centers. Example for this is big data analytics used in smart cities to optimize traffic flow, reducing fuel consumption and emissions. Networking: implementation of energy-efficient networking equipment and protocols, and use of software-defined networking to optimize resource usage. Example for this is Cisco’s energy-efficient Ethernet technology reduces power consumption during low data activity periods. Blockchain technology; transition to less energy-intensive consensus mechanisms and using blockchain for tracking and reducing carbon emissions. Example for this is the transition of Ethereum from Proof-of-Work (PoW) to Proof-of-Stake (PoS) significantly lowers energy consumption. Artificial Intelligence: development of energy-efficient AI algorithms and using AI to optimize energy consumption in various industries. Example for this is AI algorithms used by DeepMind to optimize cooling in Google’s data centers, resulting in a 40% reduction in energy usage.

4. Research Discussion

4.1. Cloud Computing

This technology has demonstrated its effectiveness as a facilitator for a wide range of IT services (Ali et al., 2015). The increasing demand for cloud-based IT services and applications highlights the necessity for advanced data centers capable of accommodating multiple web servers, storage units, and network devices, collectively referred to as CDCs (Tso et al., 2014; Gutierrez-Estevez & Luo, 2015; Tso et al., 2016). These CDCs offer a variety of services ranging from high-performance computing to extensive data analytics for end-users (Shuja et al., 2017). These large-scale CDCs are strategically located in various zones to ensure consistent distribution, meet user requirements, and account for a significant portion of total IT energy consumption (Wierman et al., 2014; Shuja et al., 2016; Koot & Wijnhoven, 2021).

Furthermore, there is a shift in IT services from single-server operations to rack-mounted blade servers. This evolution in server design has increased electronic densities, power consumption, and subsequently, heat dissipation (Ebrahimi et al., 2014; Shuja et al., 2016; Moazamigoodarzi et al., 2020). It is noted that an idle server can consume up to 70% of the power consumed by a server running at full CPU speed (Arroba et al., 2014; Lin et al., 2018; Jin et al., 2020). Additionally, many servers within data centers are underutilized or not used at all, with studies estimating that around 10% of servers in data centers were never utilized (Varasteh & Goudarzi, 2015).

The utilization of direct energy and cooling energy in CDCs contributes to an increasing demand for such centers (Davies et al., 2016; Khalaj et al., 2017). Approaches towards environmentally friendly or "green" operations of CDCs can be broadly categorized into four groups: resource management through virtualization, sustainability through renewable energy and waste heat utilization, and resource scheduling using state-of-the-art evolutionary algorithms (Patel et al., 2015; Shailendra & Singh, 2016; Haghighi et al., 2019; Oró et al., 2015; Lykou et al., 2018; Meena et al., 2016; Jangiti & Sriram, 2018). Virtualization plays a crucial role in consolidating CDC resources, allowing for efficient resource management and energy optimization (Nam et al., 2017). This virtualization layer facilitates the organization and integration of CDC resources through various techniques such as resource migration and snapshotting (Shuja et al., 2017).

Sustainability initiatives in green computing aim to reduce the environmental impact of CDC operations, with efforts towards zero greenhouse gas emissions through the adoption of renewable energy sources (Zheng et al., 2020; Cao et al., 2022). The reusability of resources and efficient management of cooling systems in CDCs are significant goals in green computing practices (Prakash et al., 2016). Strategies such as utilizing heat emissions for other purposes, like cooling processes or district heating, are being explored to enhance the sustainability of CDC operations (Nada et al., 2017; Wan et al., 2018; Ljungdahl et al., 2022).

In conclusion, cloud data centers aim to provide IT services through an optimized pay-as-you-go model (Ali et al., 2021). Green cloud computing solutions focus on energy-efficient resource allocation and task optimization, often employing graph and tree-based structures to represent network, processor, and storage devices, as well as user tasks (Madni et al., 2016). Optimization models for cloud services prioritize task completion time and cost reduction while addressing energy consumption concerns (Samriya & Kumar, 2020). The challenge lies in balancing task completion time and energy consumption in multi-resource CDCs, which requires sophisticated optimization models and the utilization of evolutionary algorithms to find optimal solutions (Peng et al., 2020; Subbaraj et al., 2021).

4.2. Mobile Computing

Modern smartphones are equipped with substantial storage capacity and computational capabilities to handle resource-intensive tasks (Dong et al., 2019), reducing the reliance on stationary/desktop servers for computation (Xu et al., 2015; Tiwary et al., 2018). This advancement has increased the resource demands of smartphones (Torous & Roberts, 2017). The emergence of sensory-rich and media-rich smartphone tools and applications often utilizes features like GPS, accelerometers, and wireless radios to provide context-aware services, leading to increased operational, communicative, and energy requirements (Shuja et al., 2017). To address the trade-off between energy consumption and output, there is a pressing need for the development of energy-efficient systems (Chen et al., 2019; Kumar & Naik, 2021). Forecasting energy usage can help in designing efficient smartphone applications and system frameworks, enabling the identification of problematic applications and maintaining system integrity (Aslam et al., 2015; Alsamhi et al., 2021).

Enhancements in the design of smartphone hardware components can contribute to energy efficiency (Ganatra et al., 2018). The structural composition of the hardware components embedded in smartphones, such as Complementary Metal-Oxide-Semiconductor (CMOS), plays a crucial role in power consumption, comprising static and dynamic power consumption (Madhu et al., 2017; Lakshmi et al., 2018). Static power consumption is device-specific and pertains to the power consumed when a transistor is not in a switching state, while dynamic power consumption relates to power consumption during transitions between logic states (Reuben et al., 2017; Mikulics et al., 2020). Techniques like power gating and dynamic frequency scaling (DFS) are employed to manage power consumption in components like CPUs, optimizing energy usage while considering performance trade-offs (Sidartha et al., 2019; Shirvani et al., 2020). Addressing components that continue to consume power despite task completion, such as Wi-Fi and GPS functions, is crucial for extending battery life and efficient smartphone operation (Geng et al., 2016; Cruz & Abreu, 2017). Various software tools, like E-prof, are available to monitor smartphone energy consumption at a detailed level, influencing overall energy usage due to their profiling activities (Shuja et al., 2017).

Software-based green computing solutions play a significant role in managing the energy budget of smartphones, including mobile cloud computing for computational offloading, energy bug handling, energy monitoring, and optimization strategies (Pramanik et al., 2019; Mansour et al., 2021). These solutions enable smartphones to connect energy-intensive functions to remote cloud servers, enhancing device longevity and efficiency (Wu et al., 2020; Maray & Shuja, 2022). Energy bug detection and resolution are essential to address abnormal power consumption issues caused by defective components or application errors (Cruz & Abreu, 2019). Software tools like Automatic Detector of Energy Leaks (ADEL) are employed to identify energy bugs and control power inefficiencies in applications, placing additional responsibilities on programmers to deploy power management measures and ensure efficient operations (Georgiou et al., 2019; Kortbeek et al., 2020).

Smartphone energy estimation serves as the foundation for green computing practices, providing valuable feedback to developers for energy-conscious application development (Farooq et al., 2019). Various methods, such as component power modeling and code analysis, are used for estimating energy consumption in smartphone applications, utilizing techniques like state-of-charge prediction and base-cost energy analysis (Mehrotra et al., 2021). These approaches help in forecasting energy consumption patterns and guiding hardware or software improvements for achieving green smartphone operations (Pereira et al., 2017). The intricacies of energy consumption reduction through efficient code design and resource optimization techniques play a key role in maintaining sustainable smartphone operations (Vijayaraghavan et al., 2017).

4.3. Internet of Things

Communication among various electronic devices without human and computer intervention is largely facilitated by the latest IoT technology (Kumar et al., 2019). Green IoT refers to a series of protocols implemented by IoT to enhance hardware and software efficiency practices (Arshad et al., 2017; Albreem et al., 2021), focusing on improving energy efficiency to reduce the environmental impact of modern services and applications (Poongodi et al., 2020; Tupe et al., 2022). Green IoT strives for eco-friendly operations through green production, redesign, and recycling/disposal practices (Waltho et al., 2019). The deployment of IoT technology involves enabling technologies, communication strategies, and embedded protocols, necessitating a discussion on benchmarking communication technologies and strategies in the context of green IoT.

4.3.1. Green Radio-Frequency Identification

Radio Frequency Identification (RFID) is a key enabler of IoT, consisting of RFID tags and readers (Flanagan & McGovern, 2022; Asci et al., 2020). RFID tags, equipped with radio-enabled microchips, act as transceivers, each bearing a unique ID and storing contextual data. These tags respond to queries from RFID tag readers, with a transmission range typically within a few meters and utilizing frequencies ranging from 124-135 kHz to ultra-high frequencies of 860-960 MHz (Muzamane & Liu, 2021). RFID tags are available in active and passive types, with active tags powered by embedded batteries and passive tags deriving power from reader signals.

To achieve green RFID practices, reducing tag size is essential to enhance recyclability, minimize non-degradable material use, and introduce sustainable options like printable, paper-based, and biodegradable RFID tags (Nowakowski, 2018; Sen et al., 2019). Energy-efficient communication algorithms and protocols are crucial for green IoT, focusing on dynamic power transmission adjustments, optimized tag estimation, and mitigating tag conflicts and unnecessary data receipt to conserve energy (Farhan et al., 2021; Popli et al., 2022; Benhamaid et al., 2022).

4.3.2. Green Wireless Sensor Network

Wireless Sensor Networks (WSNs) consist of resource-constrained nodes with limited computing, storage, and power capabilities, connected to a central sink node (Osanaiye et al., 2018; Almobaideen et al., 2019). These sensor nodes are equipped with sensors that monitor environmental conditions like humidity, temperature, and acceleration. IEEE 802.15.4 standard-based commercial WSN products employ energy-saving practices such as sleep mode activation, wireless charging mechanisms, radio optimization, and power-efficient routing strategies for green WSN operation (Gomes et al., 2017; Galmés & Escolar, 2018; Alsamhi et al., 2019; Gurusamy & Abas, 2020).

Cluster heads gather sensor data and transmit it to the sink node, leading to increased energy consumption near the sink due to heavy data aggregation. Optimizing energy usage involves periodic reporting, avoiding continuous monitoring to conserve energy and efficient content synchronization among nodes to reduce energy spikes (Yuan et al., 2021). Timely reporting intervals can minimize energy spikes associated with event-driven data reporting without the need for frequent message broadcasts to synchronize sensors.

4.3.3. Green Machine-to-Machine Communication

Machine-to-Machine (M2M) communication plays a crucial role in IoT, involving M2M and network domains for data aggregation and transmission to base stations (Shah & Yaqoob, 2016; Romeo et al., 2020). Energy consumption is a significant challenge in M2M interactions, prompting the need for energy-efficient practices such as optimized transmission power management, energy-efficient routing, intelligent task scheduling, and energy harvesting protocols (Xia et al., 2017; Chaudhari et al., 2020; Twayej & Al-Raweshidy, 2017; Dogra et al., 2022). (c) scheduling tasks on machine domain (Ali et al., 2016; Shah & Narmavala, 2018), and (d) banking on energy harvesting protocols (Yang et al., 2017; Zhou et al., 2019).

4.4. Big Data Analytics

According to Sandhu (2021), the era of big data introduces new challenges, such as rapidly expanding petabyte-scale structured and unstructured datasets with diverse forms. Achieving value optimization for decision-making in big data analytics faces major hurdles in terms of quick data retrieval and accurate search within vast data pools (Ahmed et al., 2017; Wang et al., 2018). Traditional data management systems struggle to store, handle, and analyze big data effectively, leading to the emergence of NoSQL technology to offer tailored solutions for efficient data retrieval and processing (Wang et al., 2018). Given the substantial size of data packets in big data analytics, green and environmentally friendly protocols are essential to compute, scale, and store data efficiently, requiring ample main memory availability and rapid communication mediums accessible through local physical or enterprise cloud servers (Hashem et al., 2016). Thus, ensuring resource efficiency, minimal energy consumption, and scalability of infrastructure is crucial for sustainable green big data analytics (Beneventi et al., 2017; Ganesan et al., 2020).

The application of big data analytics demonstrates its potential to optimize resource processing and storage capabilities, enabling scalability and improved productivity (Dash et al., 2019; Valaskova et al., 2021). Fundamental prerequisites such as high storage availability, technological reliability, and consistency are pivotal for the development and infrastructure of big data technology (Kumari et al., 2018; Benzidia et al., 2021). However, energy conservation and resource optimization aspects, integral to green computing, are often underexplored in recent literature (Vrchota et al., 2020). Cloud computing, a key big data technology, offers outsourced infrastructure that minimizes physical storage space requirements, facilitates multiple user access, and supports efficient storage solutions (Netto et al., 2018; Awaysheh et al., 2021). Cloud-based interfaces reduce reliance on personal computers and promote resource sharing, conservation, and low-energy consumption for high computational processes in big data analytics (Xu et al., 2019; Khashan, 2021). Cloud computing significantly contributes to enabling green big data analytics by enhancing energy efficiency and optimizing resource utilization (Yassine et al., 2019; Lv et al., 2021).

The move towards green big data analytics is evident through innovations like Green-Plum and Green-Hadoop, offering sustainable computing solutions for big data analytics (Miao et al., 2020; Shuja et al., 2017; Wu et al., 2018). Green-Plum, an open-source data warehouse under Apache Inc., facilitates rapid analytics on large-scale datasets through parallel processing and optimized query handling (Jin et al., 2020). Its cost-based query optimization feature ensures efficient data processing, enabling the analysis of extensive datasets with optimal resource utilization (Lyu et al., 2021). Green-Hadoop emphasizes the use of renewable energy sources to balance energy demand and supply for big data analytics (Cheng et al., 2015). This technology integrates solar energy forecasts and grid resources to optimize energy usage by aligning MapReduce jobs accordingly (Shuja et al., 2017). Green-Hadoop supports both batch and interactive processing modes, while real-time energy forecasting is based on historical data center workload trends (Tryfonos et al., 2018).

In sum, green big data analytics leverage cloud computing technology to maximize resource efficiency while minimizing energy consumption (Stergiou et al., 2018). Big data serves critical analytical needs across various sectors such as social networks, healthcare, industry, commerce, and corporate enterprises, facilitating prompt decision-making through data insights (He et al., 2017). Cloud-based data centers and processors play a vital role in storing and processing large volumes of data, offering green analytics solutions (Rehman et al., 2018; Thein et al., 2020). Future projections indicate cloud computing's potential to reduce energy consumption significantly, paving the way for renewable energy technologies to become more cost-effective and efficient energy sources for sustainable big data analytics solutions (Jiang et al., 2020; Karaaslan & Gezen, 2022).

4.5. Networking

Computational networks have played a pivotal role in shaping significant advancements in human civilization over the past few decades (Jarrahi & Sawyer, 2019). With the increasing integration of IT technologies and services across various industries, these networks have become more intricate, establishing connections between devices globally (Yang & Gu, 2021). Consequently, these networks and the devices within them consume a substantial portion, nearly 10%, of the total IT energy usage (Popoola & Pranggono, 2018). Essential strategies implemented to enhance energy efficiency in these networks include employing energy-efficient protocols for routing, medium access, and hand-off tasks (Luitel & Moh, 2018; Rathore et al., 2021) as well as Adaptive Link Rate (ALR) techniques aimed at adjusting link rates and utilizing hibernation states for energy-efficient computing (Tuysuz et al., 2017; Assefa & Özkasap, 2019).

Software and virtualization technologies have significantly influenced the advancement of energy efficiency in networking technologies (Hsieh et al., 2018; Montazerolghaem et al., 2020). Software-Defined Networks (SDN) separate data and control functions of network routers using a centralized controller (Prajapati et al., 2018; Abuarqoub, 2020), indirectly impacting energy consumption. The programmable interface of SDN indirectly promotes energy-efficient network operations through resource consolidation methodologies (Hu et al., 2021). Resource optimization techniques can help identify the most energy-efficient subset of network resources and can be implemented through SDN (Son et al., 2017; Kiran et al., 2020). In addition to virtualization and SDN enabling technologies, server and network resource management techniques are also utilized (Afolabi et al., 2018). SDN at the network level can facilitate the implementation of green computing policies on its programmable control plane (Liang et al., 2017). Moreover, the security functions enabled by SDN network devices can contribute to reducing operational costs and energy consumption (Son & Buyya, 2018; Chaudhary et al., 2018).

Another notable technological advancement in telecommunications is Network Function Virtualization (NFV) (Yousaf et al., 2017), which separates network forwarding and routing functions from underlying physical systems (Alam et al., 2021). NFV allows network functions like Firewalls to be executed in software form (virtual network function) on standard servers, enhancing flexibility and cost-efficiency in various IT services, particularly cloud computing services (Taleb et al., 2017). The decoupling of networking functions from physical devices through virtualization enables dynamic resource scheduling, enhancing energy efficiency (Tao et al., 2017; Sun et al., 2020). Several studies have observed energy savings and improved efficiency with NFV-based networking systems in comparison to traditional networks (Ma et al., 2018; Toosi et al., 2019; Chen, 2020). Achieving a balance between network function performance and energy efficiency is essential in the context of virtualization.

While Software Defined Networking (SDN) and Network Function Virtualization (NFV) are still evolving technologies (Yousaf et al., 2017; Kaur et al., 2020), there is a critical need for further research and development in green computing architectures utilizing these technologies, with promising possibilities for integration with other emerging technologies.

4.6. Blockchain Technology

Blockchain, also known as distributed ledger technology, represents an innovative method of data storage in which all data is shared among network participants (Jaradat et al., 2022). These participants are represented by network nodes that store, verify, transfer, and communicate data (Huang et al., 2020). Through these nodes, each participant has access to the ledger, encompassing the full history of network transactions (Dai & Vasarhelyi, 2017; Kewell et al., 2017). Nodes operate under the consensus rules of the blockchain network, establishing consensus among the network nodes when a new transaction complies with network rules and is validated against the ledger history (Bano et al., 2017). In blockchain systems, valid transactions are aggregated into a "block" and cryptographically linked to the existing chronological "chain" of blocks in the ledger (Schletz et al., 2020). This combination of cryptography, timestamping, and hashing ensures the ledger's history is tamper-resistant and highly secure (Andoni et al., 2019; Cong & He, 2019).

The rapid advancements in digital technologies and growing consumer interest in energy efficiency present significant opportunities to transition to a low-carbon energy system. Recently, distributed ledger technologies, particularly blockchain, have gained substantial attention for enhancing the security, transparency, and sustainability of the energy sector. Blockchain has been extensively discussed in the context of peer-to-peer (P2P) energy trading, with numerous articles and early product offerings exploring this concept (Andoni et al., 2019; Alam et al., 2020; Zhang et al., 2017). However, our review indicates that blockchain's application to energy efficiency remains largely conceptual and unexplored, with limited literature on the subject. This section aims to highlight key works where the concept has been discussed.

Energy Service Companies (ESCOs) are investigating potential blockchain applications to simplify energy performance contracting (EPC) (Rogers, 2018a). The prevalence of EPC business models has significantly increased in recent years (IEA, 2019), becoming a popular method for improving building energy efficiency. The EPC model involves multiple stakeholders who maintain their own records of energy baseline data, technology implementation costs, project expenses, and energy savings, leading to potential disputes when payments are due. The smart contract feature of blockchain can significantly reduce transaction costs, enabling ESCOs to undertake smaller projects by lowering the time and costs associated with setting up and managing each EPC (Rogers, 2018a). This could increase the number of ESCO projects and the total energy savings achieved. A study by Gürcan et al. developed a prototype where blockchain technology was applied to EPC, eliminating the need for third-party auditors to verify large volumes of baseline and actual consumption data (Gürcan et al., 2018). Ethan (Rogers, 2018b) has discussed how blockchain could enhance the valuation of energy efficiency by encrypting and sharing energy savings over a blockchain platform, thus improving market transparency, information security, and service reliability. The American Council for an Energy-Efficient Economy (ACEEE) is also exploring how blockchain can improve the measurement and verification process and facilitate a more dynamic energy market (Rogers, 2018b). Kawabata (2018) has examined how blockchain technology could enhance supply-side energy efficiency and address energy poverty through a decentralized energy system.

In the realm of energy-related certification schemes, a recent study explored the potential of using blockchain to trade Guarantee of Origin (GoO) certificates, also known as green certificates (Castellanos et al., 2017). These certificates, issued by regulators to renewable energy generators for each megawatt-hour (MWh) of certified renewable energy produced, record details such as when, where, how, and by whom the energy was generated, as well as ownership of the associated green assets. Although the process of transacting these certificates is cumbersome and opaque, the study by Castellanos et al. demonstrated that blockchain can ensure the authenticity of green certificates, enhance system transparency, and reduce transaction costs by eliminating the need for a third-party regulator to manage the scheme (Castellanos et al., 2017).

Energy efficiency holds substantial potential for delivering positive impacts. According to the International Energy Agency's Energy Efficiency 2018 report, improving energy efficiency could reduce consumer energy bills by over $500 billion annually (International Energy Agency, 2018). As blockchain technology evolves, it could enable consumers to trade their excess energy, providing additional incentives for energy savings and home energy efficiency improvements. As the energy efficiency market grows, blockchain technology could significantly enhance administrative processes, transparency, costs, and trust among stakeholders. Key benefits include:

• Encryption of energy savingsx: Encrypting energy savings data and sharing it over a blockchain can secure the energy efficiency market by preventing unauthorized access. Energy baseline and savings data are crucial assets, underpinning various transactions from bank payments to fees for energy service companies and technology providers. Securing this data is critical in today's digitalized world, and blockchain offers a solution for protecting customer energy savings data.
• Exchange of energy savings: Beyond P2P energy trading of excess electricity generation, blockchain technology holds potential for trading energy savings within local communities. Energy savings data could be encrypted and stored on a blockchain platform, enabling transactions for balancing energy bills or purchasing additional energy services.

4.7. Artificial Intelligence

Advances in sensing devices have enabled a shift towards the fourth industrial revolution (Gooneratne et al., 2020). The large volume of data produced by these devices is pushing the technology front of a new generation of AI for IoTs applications (Misra et al., 2022). These applications are expected to make important decisions in the real world instantaneously rather than offloading data to the cloud servers (Javaid et al., 2018; Wu et al., 2019; Zhao et al., 2022). Such a step change in technology requires significant strides in energy efficiency, which continues to be a primary design challenge for IoT hardware designers (Biswas & Chandrakasan, 2018; Neftci, 2018).

Existing AI systems predominantly follow the principle of neural networks (NNs). Originally inspired by Rosenblatt’s neural automaton in 1957 (Lei et al., 2020), modern NNs have evolved in complexity across different application domains (Boutaba et al., 2018). Typically, NNs define a learning problem by finding the weighted sum of all inputs in the training phase, organized in multiple layers (Azzouni et al., 2017). The weight updates are defined by a normalized activation function and are performed through rigorous gradient descent exercises. When implemented in hardware, the modular electronic neurons require arithmetic-heavy circuits, such as multiply–accumulate (MAC) units (Yaseen et al., 2019; Lei et al., 2020). The number of these units can quickly grow with more inputs and added complexity of the learning problem (Benini, 2017). Given such a scale of arithmetic complexity, achieving required energy efficiency and performance in NNs can be daunting, which is exacerbated further by the large volume of data generated by IoT devices (Wang et al., 2019; Khan et al., 2020).

Over the last two decades, significant progress has been made in energy-efficient NN hardware research (Ding et al., 2018). A vast majority of the existing work have considered pruning arithmetic complexity to save energy by exploiting the natural resilience of AI applications to minor deviations or error (Li et al., 2019; Armeniakos et al., 2022). Examples include precision scaling (Wang et al., 2019), approximate logic designs (Shafik et al., 2018; Qiqieh et al., 2018), new analogue or mixed-signal circuit designs and hardware/software co-design for NNs (Zhong et al., 2018; Lei et al., 2020). Recently, there have been overwhelming interests in moving away from arithmetics to using binary logic as the core building blocks (Fyrbiak et al., 2018). Binarized neural networks (BNNs) are an example of this development (Nurvitadhi et al., 2016). The key goal is to condense advanced AI workloads with low-energy footprints. However, this can make the learning process (i.e. accuracy and convergence) sensitive to how gradient descent is designed, which is still arithmetic based (Mohammadi et al., 2018; Zhang et al., 2022).

Recently, the Tsetlin machine has been proposed as a promising machine learning (ML) algorithm based on learning automata. The Tsetlin machine simplifies the traditional learning automata by discrete-step action updates through Tsetlin automata, defined as the finite automata with linear tactics. For action updates, each Tsetlin automaton uses rewards for reinforcing an action and penalties for weakening the automaton confidence in performing the action. This discretization with linear step updates allows for formulating the learning problem using powerful propositional logic (Lei et al., 2020); furthermore, it simplifies the learning mechanism, enabling efficient on-chip learning. The input data in a Tsetlin machine are encoded in binarized form as a set of propositional logic variables, called literals. These literals are used to build the logic expressions corresponding to inference classes through ensembles of parallel Tsetlin automata, called clauses, during training (Shafik et al., 2020). When training is completed, the inference outputs are described by binarized classifications.

The logic-based structure of Tsetlin machines provides opportunities for energy-efficient AI hardware design. This will require addressing the major challenges of the systematic architecture allocation of low-level resources as well as parametric tuning and data binarization, which cannot be achieved by using high-level synthesis or hardware-assisted acceleration tools.

5. Research Implications

This systematic literature review leads us to highlight several key implications for future research for the greening of IT technologies.

5.1. Implications to Theory

This research delves into a comprehensive analysis of energy consumption sources and greenhouse gas emissions stemming from various IT technologies. A significant aspect of this study is to enhance our comprehension of the energy-saving potential inherent in each technology, as well as to explore strategies for cross-technology adoption. By elucidating the intricate interconnections and interdependencies among different IT technologies, this paper sheds light on the significance of informed decision-making in balancing high-performance expectations with the imperative to reduce resource consumption, streamline communication, and simplify computing operations. The foundational discourse on emerging IT technologies and the imperative for sustainability underscores the necessity of strategic trade-offs to optimize environmental outcomes without compromising operational efficiency.

Moreover, as illustrated in the provided theoretical framework in Table 2, a structured approach is delineated to identify key energy-saving imperatives within IT technologies. The framework serves as a foundational guide for categorizing IT technologies into essential components, facilitating a granular exploration of techniques or algorithms aimed at mitigating power consumption within each category. This structured framework offers a roadmap for researchers and practitioners to navigate the complexities of IT technologies, enabling targeted strategies for maximizing energy efficiency across diverse technological domains.

Furthermore, the convergence of cutting-edge technologies necessitates a nuanced understanding of how IT solutions can seamlessly integrate to drive sustainability and energy efficiency. Building upon this foundational framework, future research should explore innovative approaches for leveraging synergies between different IT technologies to achieve holistic greening goals. Acknowledging the evolving landscape of digital transformation and environmental responsibility, a comprehensive framework incorporating a blend of technological solutions, adaptive methodologies, and cross-domain applications is imperative for optimizing greening initiatives in IT ecosystems.

Moreover, the theoretical underpinnings and practical implications of this study extend beyond energy reduction strategies to encompass broader sustainability objectives. Embracing a holistic perspective on IT greening entails not only enhancing energy efficiency but also promoting sustainable practices, minimizing carbon footprints, and fostering eco-friendly technologies. By synthesizing theoretical insights with empirical findings, researchers can inform the design and implementation of tailored greening strategies that resonate with the evolving needs of the digital landscape.

In essence, the theoretical implications outlined in this study serve as a steppingstone for advancing knowledge in IT greening and sustainable technology practices. By fostering a multidisciplinary approach that integrates theoretical frameworks with practical applications, researchers can propel the discourse on greening IT technologies forward, paving the way for transformative solutions that harmonize technological innovation with environmental stewardship.

5.2. Implications to Research

This study underscores the critical need for a SR to synthesize research on emerging IT technologies within consolidated optimization models to enhance environmental sustainability. Existing literature offers a diverse array of theoretical perspectives on optimizing energy efficiency within specific IT technologies and their components, such as software components or communication layers. Many IT-based applications span multiple technological domains, for instance, a smartphone app functioning on a mobile device interconnected to a cloud server, activating various IT technologies like mobile computing, cloud computing, networking, IoT, and WSNs. Notably, current energy estimation techniques often overlook external factors contributing to energy consumption. Conventional energy-saving strategies often adopt a segmented approach, inadvertently overlooking the interconnectedness of different technological layers, user demands for high-quality service, and potential trade-offs between efficiency gains in one area leading to added consumption in another. Delving deeper, the intricate interplay of IT technologies, varying software and hardware components, and performance requirements necessitate a holistic approach to energy optimization strategies for sustainable outcomes. Future research should explore comprehensive energy-saving frameworks that account for the synergies and trade-offs among diverse IT technologies.

Expanding on this, another significant research avenue lies in exploring the transformative potential of IT technologies as tools for energy conservation and reducing carbon footprints across various industries. For instance, the strategic deployment of wireless sensors across power grid networks, despite the initial surge in energy consumption, presents substantial benefits. This system equips electricity providers with actionable insights to optimize power generation and distribution, fostering greater energy efficiency and environmental sustainability. It is crucial to consider the holistic impact of greening initiatives in IT technologies on broader industries, recognizing the intricate dynamics between increased energy consumption for data processing and communication demands vis-a-vis overarching environmental benefits. Researchers are encouraged to develop optimization frameworks that encompass the ripple effects of energy usage changes on diverse sectors, ensuring a comprehensive assessment of the overall greening achievements. While achieving greener outcomes may involve transient spikes in energy consumption due to data processing and communication demands, a systemic approach considering the net environmental implications for various industries is vital to attain optimal and sustainable results.

In essence, the academic community is urged to delve deeper into energy-saving frameworks that embrace the complex interactions among diverse IT technologies and industries, elucidating the interconnected dynamics between energy consumption and greening efforts. By exploring these multifaceted interdependencies, researchers can pave the way for innovative and sustainable IT solutions that drive environmental preservation and resource optimization across broader industrial landscapes.

Figure 5. Cloud network – power grid interconnection.

Figure 5. Cloud network – power grid interconnection.

5.3. Implications to Practice

This review study presents crucial practical implications essential for the effective and expedited implementation of greening initiatives within the IT sector. Estimates indicating that the IT industry accounts for approximately 2.4% of global electricity consumption, coupled with its rapid expansion, underscore the significant potential for power savings through the adoption of greening practices (Koot & Wijnhoven, 2021). The discussion highlights the pivotal role of cloud computing for advancing greening efforts for two primary reasons. Firstly, the expansive scale of the cloud infrastructure makes greening techniques highly impactful (Ali et al., 2020). Secondly, the integration of cloud computing across various IT domains, such as mobile devices leveraging cloud processing capabilities for high-energy operations, underscores its integral role. Technologies like IoT, WSNs, and big data analytics leverage cloud services for data collection, storage, and processing, realizing efficiencies and reducing reliance on personal computers and local servers. Thus, enhancing the sustainability of data centers not only benefits their operations but also cascades into greening other IT technologies.

The integration and widespread adoption of renewable energy sources emerge as a cornerstone in fostering a sustainable environment (Jamal & Khan, 2020). Research indicates that the IT industry has the potential to reduce energy consumption significantly, up to 64%, by embracing renewable energy technologies (Al-Jarrah et al., 2015; Fais et al., 2016). Contemporary research focuses on addressing challenges associated with renewables, such as cost and energy generation unpredictability, offering innovative solutions for wider adoption (Osman et al., 2023). Collaboration among IT companies, organizations, and governmental bodies is imperative in driving the broad deployment of renewable energy sources. Governmental entities and organizations need to establish clear goals, with industry stakeholders working collectively to achieve these targets. Cloud service providers and renewable energy firms stand to benefit from this collaboration, with cloud data centers transitioning to clean energy sources. The symbiotic relationship between renewables and cloud computing facilitates better networking of geographically dispersed renewable energy resources, enabling efficient management, evaluation of performance, maintenance, and expansion plans.

Furthermore, the onus is on software development in driving greening efforts within the IT sector. Simply focusing on providing functionality and rapid response times in software is no longer sufficient. Modern software developers are mandated to possess a deeper understanding of energy consumption patterns and efficiency considerations (Alvi, 2017). Effective software design with energy efficiency as a core objective is pivotal to synergize with energy-aware hardware technologies and advance sustainable IT practices. Incorporating energy analysis models, tools, and energy-aware compilers into the software development process is crucial to optimize energy consumption. Additionally, the development of standardized testing utilities for profiling and benchmarking software applications will enhance the energy efficiency assessment of published software, further aligning software development practices with green IT objectives.

Moreover, the convergence of cloud computing with renewable energy sources presents a prime opportunity to accelerate green initiatives within the IT sector. By leveraging renewable energy and cloud services in tandem, stakeholders can achieve a more sustainable and environmentally conscious ecosystem. This collaborative approach not only enhances the energy efficiency of data centers but also sets a precedent for holistic greening efforts across IT domains. The alignment of cloud technologies with renewable energy sources signifies a pivotal step towards achieving greener IT operations and fostering an eco-friendly digital landscape.

In tandem with the adoption of renewable energy and cloud computing solutions, ongoing advancements in software development practices hold the key to unlocking greater energy efficiency in IT operations. Modern software design must prioritize energy-conscious coding practices, aiming to optimize energy usage and minimize environmental impact. Integrating energy analysis tools and energy-aware compilers into the software development lifecycle empowers developers to infuse energy efficiency considerations into software design, maximizing the sustainability benefits of IT technologies.

In conclusion, a collaborative and multi-faceted approach involving cloud computing, renewable energy sources, and energy-efficient software development practices is pivotal in steering the IT industry toward a greener and more sustainable future. As stakeholders across the IT landscape embrace these practices, they can collectively drive impactful changes towards energy efficiency, reduced carbon footprints, and a more environmentally conscious approach to technology innovation. By fostering partnerships, investing in renewable energy technologies, and prioritizing energy-efficient software design, the IT sector can play a pivotal role in advancing environmental sustainability and greener IT practices at a global scale.

6. Research Limitation

This research has two limitations, the first one despite the deployment of a thorough search strategy, some studies on the different advanced technologies that have been presented in this review study were not included, such as gray literature and reports that were not published in the selected databases that were reviewed. The second notable limitation of this systematic review is the scarcity of long-term data and comprehensive assessments of the environmental impact of greening emerging IT technologies. While the literature provides insights into the immediate effects and short-term benefits of adopting green IT practices, there is a gap in understanding the extended, downstream consequences. This limitation arises from the relatively recent adoption of many green IT initiatives, making it challenging to assess their full ecological implications over extended periods. Therefore, the review primarily captures a snapshot of the current state of knowledge, and further research is required to monitor and evaluate the sustainability of green IT solutions over the long term. Longitudinal studies and extended monitoring are essential to gain a more comprehensive understanding of the lasting benefits and potential trade-offs associated with green IT adoption. Additionally, assessing the scalability and applicability of green IT practices across diverse IT ecosystems remains a valuable area for future research. This research limitation highlights the need for more in-depth, long-term studies to fully comprehend the lasting impact of green IT technologies and their scalability in various contexts.

7. Conclusion

The increasing integration and demand for IT technologies are driving higher energy consumption, prompting a global need for more environmentally friendly practices within the industry. In response, researchers and IT professionals have developed various algorithms and protocols to promote sustainability. These strategies include enabling low-power states during idle periods, making energy-conscious decisions, and optimizing resource allocation. However, it's essential to recognize that reducing energy consumption can impact system performance. The extent of energy conservation depends significantly on the specific application's usage, and extremely aggressive energy-saving measures can affect the system's longevity due to frequent power cycling.

Addressing the environmental impact of emerging IT technologies requires a collaborative effort from governments, industries, researchers, and consumers. This endeavor is not solely technological but represents a societal commitment to sustainability. The challenge is substantial, yet the potential for innovation and positive change is immense. By adhering to the key requirements outlined in this review, the IT industry can significantly contribute to a more sustainable and environmentally responsible future. The findings and recommendations presented here are intended to guide further research, fostering the transformation of emerging IT technologies into energy-efficient and sustainable solutions. Through these efforts, the IT sector can play a pivotal role in advancing global environmental sustainability.