Localisation of Room Occupants in Home Environments using Unobtrusive Sensing Solutions

This paper proposes the localisation of room occupants in home environments using Unobtrusive Sensing Solutions (USSs). The ability to localise room occupants in home environments can help in the objective monitoring of sedentary behaviour, control of appliances and temperature regulation inside buildings. While wearable sensors can provide tangible information on health and wellness, they have battery-life issues and the inability to perform prolonged monitoring. This work uses heterogeneous USSs in the form of an Infrared Thermopile Array (ITA-64) thermal sensor and a Multi-Chirp Frequency Modulated Continuous Wave Mono-pulse (MC-FMCW-M) Radar sensor to monitor room occupants. Digital filters, interpolation and background subtraction algorithms were used to process the thermal images gleaned from the ITA-64 thermal sensors. The MC-FMCWM Radar sensor used multi-chirp and Doppler shift principles to estimate the exact location of the targeted room occupants. The estimated distances from the Radar sensor were compared with ground-truth values. Experimental results demonstrated the ability to identify thermal blobs of occupants present in the room at any particular time. Data analyses indicated no significant difference (p = 0.975) and a very strong positive correlation (r = 0.998) between the ground-truth distance values and those obtained from the Radar sensor.


Introduction
The history of wearable devices dates to the 15th century with the advent of simple wristwatches to tell time [1]. In the past five decades, additional functions have been integrated into Wearable Sensing Solutions. These add-ons have included health monitoring [2], [3] and activity recognition functions [4]- [6], amongst others. While emphasising the numerous advantages of using wearable devices for monitoring, it is worth highlighting that some of the devices are designed with unpleasant entanglements. Others are recognised for having battery-life and wearability issues. As an example, a study by [3] on localisation for activity recognition required participants to wear sensors on seven different parts of the body. With these numerous sensors and entanglements, users can be discouraged from wearing and continual usage of the proposed technology. Also, uninterrupted usage of wearable devices for localisation in a home environment can be impracticable, uncomfortable or undesirable due to wearability issues and other issues such as muscle artefacts, which can affect the quality of data obtained from the devices. In addition, the ability of users to remember to charge and wear devices can result in frequent interruptions of data acquisition processes.

Related Work
Unobtrusive Sensing Solutions (USSs) are being used increasingly for the purposes of indoor and outdoor localisation. They include monitoring devices that are not worn or held on any part of the human body. Recently, these devices have been applied in many Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 16 August 2021 doi:10.20944/preprints202108.0333.v1 health-related and other indoor settings, such as monitoring at-risk individuals [7], the performance of Activity of Daily Living (ADLs) [8], exercise rehabilitation [9] and localisation frameworks [10], [11]. Whilst some of these sensing solutions can work independently, others need to be integrated with wearable solutions.
The unmanned aerial vehicle is an example of an outdoor system for manoeuvring rugged terrains in the security and surveillance framework. Another example is the global positioning system (GPS) which can be used to collect data for location estimation and mapping. Incident localisation and identification such as earthquakes, highway congestions, environmental pollution, missiles and drones invasion can also be carried out with the help of the GPS, global navigation satellite system and other related IoT technologies [12]- [14]. Unfortunately, GPS is not suitable for indoor data acquisition in places such as underground buildings, railway stations and parking garages [15]. RFID systems have been widely used in IPS. These systems use radio waves to identify and track their targets. The tracking information is stored on tags placed on intended targets. RFID systems are used in many industries, such as the automobile manufacturing industries [18], to track the progress and location of items along assembly lines. At security checkpoints, RFID tags are used in many locations as a replacement for magnetic stripe cards. Nevertheless, passive tags suffer from received-signal-strength-indication loss. Furthermore, active RFID tags suffer from battery-life challenges [18]. Although efforts have been made to reduce these problems through the use of advanced RFID technologies, limitations, such as single object tracking, direction dependence and non-conformity with the Federal Communications Commission power limits standard [19], are still being researched.
Spot localisation has also been explored using information from the physical layer The Ultra-Wide-Band (UWB) Radar is a high bandwidth radio technology system that utilises low energy for short-range communications. Access points and algorithms for UWB Radar during an indoor localisation study was proposed by [23]. Mobile units (MUs) were used as targets for this study. Although a high accuracy was reported in the resulting overall measurements, the study setup posed challenges to home-based users.
Also, MUs were deployed in the study instead of human targets.
This study, therefore, proposes the localisation of room occupants in home environments using USSs such as ITA-64 thermal sensor and an MC-FMCW-M Radar sensor. The main contribution of this work are itemised as follows: (i) to determine the actual location of room occupants in home environments using heterogeneous USSs, (ii) to demonstrate the advantages of using USSs such as Radar and Thermal SS in indoor monitoring against the wearable solutions, (iii) to demonstrate the advantages of using dual and complementary monitoring solutions compared with using a single SS, (iv) to monitor human presence in a room through the usage of USSs such as thermal and Radar SS.

Materials and Methods
The experimental setting comprised a Silicon-Germanium (SiGe) based 24GHz Multi   The pictorial view of the sitting arrangement of the targeted room occupants is presented in Figure 2. The targeted room occupants referred to as O1, O2, O3, O4 and O5 (in this study) were at seats number 1, 2, 3, 4 and 5, respectively. In Figure 2( Given that the sensors were mounted on a tripod stand 1.  The Radar sensor was used to record live data from the targeted occupants for ten days. The sensor was interfaced with MATLAB software. The ITA thermal sensor recorded the thermal blobs of all the occupants simultaneously to allow for easy playback. It is important to emphasise that the participants' daily activities and office duties were neither influenced nor interrupted in any way during the data acquisition process. Also, their sitting positions were not adjusted either, and data collection continued without any adjustment to their daily routines.

Results
Data obtained from the MC-FMCW-M Radar sensor was analysed using MATLAB 2018 and Minitab 18 statistical software packages. Range values gleaned from the sensor were checked for zero-factor error (ZFE) before being exported to Minitab for descriptive analysis. A total of 565 frames were considered for descriptive analysis. The mean, standard error of the mean (SEM) and standard deviation (SD) of the range values were computed for each occupant as presented in Table 1. Other parameters considered were the variance and the coefficient of variance (CoefVar) of each of the range values. Thermal blobs gleaned from the ITA sensor were binarised to improve their granularity. A digital image interpolation algorithm was used to eliminate thermal blobs interference. Var. Results presented in Table 1   The thermal blobs of the room occupants were captured with the ITA sensor when all the occupants were at their respective seats. This was also the case with the MC-FMCW-M Radar. Unlike other studies [18] [19] that restricted participants to a predetermined pattern, the occupants (in this study) were not advised to conform to any specific pattern during data collection. This approach was taken to allow for real-life behaviour rather than experimentally constrained behaviour, which may influence experimental outcomes. Figure 5(a) and (b) present the ITA actual and binarised thermal blobs, respectively of the targeted room occupants. Acquired thermal blobs were binarised ( Figure 5(b)) to enhance clarity, more so, to represent a distinct thermal blob of each targeted participant. From Figure 5(b), O1 and O2 had almost a third of a full human thermal blob due to their proximity to the sensor.
Thermal blobs from O3, O4 and O5 appeared smaller; however, they were still visible and distinguishable. The occupant marked 'X' was not included in the study. Additionally  In Figure 6, the thermal blobs of O1 and O3, O5 and X interfered with one another.
Also, loose pixels are observed, which can be mistaken for additional human thermal

Discussion
The experimental evidence from this study has informed the benefits of using USSs This study has many areas of practical application, including monitoring ageing adults in a home environment, multiple occupants in a care facility, sedentary behaviour in a workspace [26] and temperature control in offices and homes, to name but a few. Although diverse smart SSs are currently used for these purposes, a recent systematic survey by ElHady and Provost [25] detailed that more work was still required in terms of addressing failure and faults in the existing systems.
The main challenge with this study was the ZFE. It occurred under two circumstances in this study. Firstly, when a reflected signal from a target is not received at the sensor. Secondly, when an occupant left their seat during the experiment. While the first instance has a negative influence on the data recorded by the sensor, the second instance did not affect the recorded data. In order to eliminate the first case, data were manually cleaned with the help of complementary information from the ITA sensor. This further highlights the essence/benefits of heterogeneous sensing considered within this study.

Conclusions
This study presented privacy-friendly heterogeneous USSs for localisation of room occupants in a home environment. An MC-FMCW-M Radar and an ITA SSs were considered. Each SS was non-wearable, contactless and unobtrusive. The sensors individually acquired data and processed the data for complementary functions. Data acquisition was carried out for ten days, and a total of 565 data frames were analysed from both sensors.
Experimental results clearly indicated that the Radar sensor estimated the range values of the targeted participants with high accuracy. The ITA sensor complementarily recorded distinct thermal blobs, which, in combination with the Radar sensor data, estimated the actual sitting location of the targeted occupants. These traditional roles helped to avoid zero-factor and false-positive errors. Future work will attempt the use of these USSs alongside data mining algorithms to localise and predict the speed at which ADLs are performed in a home environment.
Funding: Research is funded by the EU's INTERREG VA program, managed by the Special EU Program Body (SEUPB).