1. Introduction

2. Related Work

3. Classification of jamming Attacks

4. Jamming Detection and Description Scenario

4.1. Predication Location

Figure 1 supports the assumption that each vehicle equipped with an On-Board Unit (OBU) occupies one lane under normal operating conditions without electronic jamming. At time t, the Host Vehicle (HV) receives a Basic Safety Message (BSM) from the Remote Vehicle (RV) containing information such as vehicle ID, location, type, heading, speed, and acceleration. The HV generates the same set of information about its own status. Using this information, the HV can calculate the current distance (Distance (t)) between the two vehicles and the Packet Delivery Ratio (PDR) at time t (PDR (t)), as shown in Figure 1a. As time passes, the vehicles move, and new BSMs from the RV provide updated information on its actual movement. The HV can use this information to predict the future distance between the two vehicles and the corresponding PDR values at time t+Δt, as shown in Figure 1b. As time passes, the two vehicles will move to different locations, as depicted in Figure 4.1c. Consequently, new Basic Safety Messages (BSMs) will be transmitted from the Reference Vehicle (RV) to indicate the actual movement. By comparing the estimated values with the actual values derived from these BSMs, any inconsistencies can be identified, indicating abnormal behavior.

Figure 1. Jamming Detection Scenario.

Figure 1. Jamming Detection Scenario.

5. Model designed to detect jamming attacks on VANETs.

Figure 2. Model designed to detect jamming attacks.

Figure 2. Model designed to detect jamming attacks.

6. Jamming Attack Recovery Algorithm for Dedicated Short-Range Communications (DSRC) Safety Applications in Vehicular Ad-Hoc Networks (VANET).

6.1. Reliability and Redundancy

• Figure 3 showcases several vital Safety Applications, which we'll delve into below:
• Forward Collision Warning (FCW) – Pictured in Figure 3a, the FCW alerts the driver of the Host Vehicle (HV) about a potential rear-end collision with a Rear Vehicle (RV) ahead, moving in the same lane and direction. This system proves especially valuable when the HV is closing in on a vehicle that's either decelerating or has come to a halt.
• Emergency Electronic Brake Lights (EEBL) – As illustrated in Figure 3b, the EEBL is somewhat of a toned-down version of FCW. Upon receiving data that an RV ahead is braking sharply, it prompts the HV driver to slow down. The significance of this feature is amplified when an obstruction (like a big truck) limits the HV driver's view.
• Do Not Pass Warning (DNPW) – Depicted in Figure 3 c, DNPW sounds an alert to the HV driver during an overtaking maneuver if there's an incoming vehicle from the opposite direction.
• Blind Spot Warning + Lane Change Warning (BSW+LCW) – Demonstrated in Figure 3d, this Safety Application cautions the HV driver wanting to switch lanes if another vehicle—moving in the same direction—is present in their blind spot.
• It's imperative to note that the success and reliability of these DSRC Safety Applications are heavily dependent on the Basic Safety Messages (BSM) from the RV. If the HV doesn't receive these messages or they aren't frequent enough, the effectiveness of the application may be compromised.

Figure 3. Safety Application Scenarios.

Figure 3. Safety Application Scenarios.

Remember that each On-Board Unit (OBU) sends out a Basic Safety Message (BSM) every 100 ms using the safety channel (CH172). This results in a broadcast frequency of 10 BSM/s as referenced in sources [9] and [8]]. Building on this, we're looking into the idea of redundancy by amplifying this message frequency. We're keen to understand its effect on the reliability of Safety Applications and the subsequent overhead. Take, for example, the forward collision warning system illustrated in Figure 3.a. In this scenario, the rear vehicle (RV) brakes abruptly, possibly to avoid a hurdle. Such abrupt braking will trigger a brake system status notification in the BSM of the RV. From the perspective of the Host Vehicle (HV), this means it should receive at least one BSM reflecting this status update early enough to factor in the driver's reaction time. Consequently, the reliability of the Safety Application can be described as the likelihood of the HV getting a minimum of one BSM notification in time to allow the driver to react, specifically by time t_react.

To harmonize with the conventional definition of reliability – that is, R(t) representing the chance that a system operates as intended throughout a set timeframe [0, t], as mentioned in source[8] – we can define our specific application reliability, R(t)app, as the likelihood of receiving a minimum of one BSM by time t_react. If we label the BSM as BSMi, then it's imperative to receive one of the BSMi, where i ranges from 1 to x, with BSMx being the final BSM before t_react. Hence, the application would only falter if not even one BSM is received by t_react. The metric for unreliability, Q(t)app, can be represented as 1 − R(t)app. This metric essentially gives the probability of not receiving any BSMi within the range of i = 1 to x.

At a regular interval of 100 milliseconds, every On-Board Unit (OBU) transmits a Basic Safety Message (BSM) on the safety channel (CH172), resulting in a message frequency of 10 BSMs per second[26] and [9]. To assess the impact of redundancy on the dependability and overhead of Safety Applications, we increase the BSM message rate. Let us consider the forward collision warning application where the Recreational Vehicle (RV) abruptly applies the brakes to evade an obstacle, resulting in a brake system status alert in the RV's BSM. To inform the driver sufficiently in advance, allowing adequate reaction time, the Host Vehicle (HV) needs to receive at least one BSM containing this status. As a result, the reliability of the Safety Application pertains to the probability of the HV receiving at least one BSM message before the reaction time threshold, i.e., at time t_react.

Following the standard definition of reliability, where R(t) is the probability of the system performing in accordance with the specifications throughout the time interval [0, t] [27], we define our application reliability R(t)app as the probability of receiving at least one BSM message before at time t_react. Assuming the i^th BSM as BSM_i, at least one of the BSM_i, i = 1... X, must be received, where BSM_x refers to the last BSM before at time t_react. Consequently, the application fails only if no BSM message is received before t_react. The unreliability 〖Q(t)〗_(app =1- 〖R(t)〗_app ) i.e., the probability of not receiving any BSM_i, i = 1... X, can be expressed as:

$${\mathrm{Q}\left(\mathrm{t}\right)}_{\mathrm{a}\mathrm{p}\mathrm{p}=}{\prod }_{\mathrm{i}=1}^{\mathrm{x}}{\mathrm{Q}\mathrm{i}}^{\left(\mathrm{t}\mathrm{i}\right)}$$

(7)

Where $Q_i$is the probability that BSM_i was not received at t_i. It should be noted that $Q_{i(t_{i)}}$is the packet error probability of ${\mathrm{B}\mathrm{S}\mathrm{M}}_{\mathrm{i}}$. If N redundant channels are used, then

$Q_N{\left(t\right)}_{app}={\prod }_{j=1}^N{Q}_{ci}\left(t\right)$ Where $Q_{{ci}^{\left(t\right)=}}{\prod }_{i=1}^x{Q}_{i\left(t_i\right)}$ represents the unreliability for each channel, as

Equation .3, introduced in [28], can be used to calculate the Safety Application's unreliability and channel redundancy for a single channel. Increasing the BSM rate has the potential to improve the system's reliability, but the impact depends on the intensity of jamming, as illustrated in Figure 4. If the left region experiences jamming, the jamming detection will activate the Safety Application's fail-safe mode. The right region remains unaffected by jamming. However, the central region is crucial, as increasing the BSM rate can mitigate moderate jamming. This strategy is not only applicable to the safety channel (CH172) but can also be extended to other channels, such as CH178 using ACM in a dual redundant approach or incorporating CH184 using PVD for a triple redundant scheme.

Figure 4. Jamming impact regions.

Figure 4. Jamming impact regions.

6.2. Effectiveness of various BSM Rates.

The MAC layer can face additional strain with an increase in BSMs due to higher BSM rates, which may lead to a decrease in PDR as a result of collisions. Determining the upper limit of BSM rates is dependent on several critical factors, including the number of vehicles in the vicinity, data rate, and message size. To gain insight into this upper limit, Figure 5 illustrates the maximum available BSM rates for different data rates and two sample BSM sizes, 300 and 180 Bytes, using a PHY Preamble of 32μs, DIFS of 64μs, and PLCP header of 8μs. For instance, when sending a 300 Bytes message at a data rate of 6Mbps.

Figure 5. Upper bound on BSM rates for different data rates.

Figure 5. Upper bound on BSM rates for different data rates.

$\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{s}\mathrm{m}\mathrm{m}\mathrm{i}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{D}\mathrm{e}\mathrm{l}\mathrm{a}\mathrm{y}=\frac{\mathrm{M}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{a}\mathrm{g}\mathrm{e} \mathrm{S}\mathrm{i}\mathrm{z}\mathrm{e}}{\mathrm{D}\mathrm{a}\mathrm{t}\mathrm{a} \mathrm{R}\mathrm{a}\mathrm{t}\mathrm{e}}=\frac{8\times 300}{6000000}-400\mathrm{M}\mathrm{s}$ And now by adding all delays 32μs + 64μs + 8μs + 400μs, these sums up to a total delay of 504μs per message. As BSM data rates higher than 6Mbps have been shown to be too unreliable in the presence of jamming, the data rates that should be used by DSRC Safety Applications are 3Mbps and 6Mbps [65]. For a 6Mbps data rate, the maximum number of messages the media can handle is 1984 BSM/s for 300 Bytes message size.

This can be calculated considering.$\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{M}\mathrm{a}\mathrm{x}\mathrm{i}\mathrm{m}\mathrm{u}\mathrm{m} \mathrm{T}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{u}\mathrm{g}\mathrm{h}\mathrm{p}\mathrm{u}\mathrm{t}=\frac{\mathrm{M}\mathrm{a}\mathrm{s}\mathrm{s}\mathrm{a}\mathrm{g}\mathrm{e} \mathrm{S}\mathrm{i}\mathrm{z}\mathrm{e}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{l} \mathrm{D}\mathrm{e}\mathrm{l}\mathrm{a}\mathrm{y}}=\frac{2400 \mathrm{b}\mathrm{i}\mathrm{t}\mathrm{s}}{504 \mathrm{M}\mathrm{s}}=\mathrm{4,671,904} \mathrm{b}\mathrm{i}\mathrm{t}\mathrm{s}/\mathrm{S}\mathrm{e}\mathrm{c}$.

Thus, when sending a BSM of size 300 Bytes (2400 bits), the total messages that can be handled by the media is$\frac{\mathrm{M}\mathrm{a}\mathrm{x}\mathrm{i}\mathrm{m}\mathrm{u}\mathrm{m} \mathrm{T}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{u}\mathrm{g}\mathrm{h}\mathrm{p}\mathrm{u}\mathrm{t}}{\mathrm{P}\mathrm{a}\mathrm{c}\mathrm{k}\mathrm{e}\mathrm{t} \mathrm{S}\mathrm{i}\mathrm{z}}=1984\mathrm{B}\mathrm{S}\mathrm{M}/\mathrm{S}$. Likewise, at 6 Mbps and for a message.

Assuming a message size of 180 bytes, the media can handle a maximum of 2906 BSMs per second. With each vehicle sending 10 BSMs per second, one can estimate the upper limit of vehicles that the media can handle. However, it is important to note that these calculations do not take into account the potential collisions that may occur when multiple vehicles attempt to send BSMs simultaneously. Collisions can lead to packet corruption, bandwidth consumption, and a decrease in PDR, particularly as the number of vehicles increases. It is worth noting that VANET uses the DCF and CSMA/CA protocols of the IEEE 802.11 standard. Despite these considerations, the maximum number of messages shown in Figure. 4 was calculated without accounting for collisions, and it is important to consider the impact of redundant BSMs on the medium in a more realistic scenario.

Collisions in VANETs can occur through direct collisions or hidden terminals. Direct collisions happen when the sender and receiver are within each other's transmission range, and they send messages simultaneously due to similar back-off times. In contrast, hidden terminal collisions occur when three or more nodes are positioned such that the outer nodes are not within each other's transmission range, but they are within the range of the middle node. This leads to more collisions since the outer nodes cannot sense each other's presence, resulting in simultaneous communication with the middle node. To prevent these collisions, wired and wireless networks use several mechanisms, including physical carrier sensing and virtual sensing. Physical carrier sensing involves the sender monitoring the medium and deferring transmission if the medium is busy. Only when the medium is idle, the sender transmits the data frame after a random back-off time to avoid direct collisions with other nodes competing for the medium.

Physical carrier sensing involves monitoring the medium and waiting for an idle period before transmitting data to avoid direct collisions. Virtual sensing, on the other hand, sets a Network Access Vector (NAV) based on Request-to-Send and Clear-to-Send (RTS/CTS) frames. Before transmitting data, a source node sends an RTS and waits for a CTS reply from the destination, and hidden nodes outside the source range can still hear the CTS reply and set their NAV accordingly to reduce collisions in hidden terminal situations. However, virtual sensing is not suitable for safety applications in VANETs, where minimal delay is crucial for broadcasting BSMs to high-speed vehicles. Therefore, hidden terminal situations have a more severe impact on safety applications in VANETs. The impact of transmission collisions on Packet Delivery Ratio (PDR) is studied using the IEEE 802.11p MAC protocol in [29], which measures performance for both hidden and direct collision cases. The average number of vehicles in the transmission range is denoted by N_tr.

$$N_{tr }=1+2\beta R$$

(8)

Which β represents the density of vehicle [vehicles/km] and R is transmission range. The queue utilization ρ can be expressed as

$$\rho =\lambda E\left[S\right]$$

(9)

Where λ is the packet generation rate [packets/sec] and E[S] is the average of the servant time.

Now let as τ be a probability of that a transmit vehicle in the slot random considering which a packet in a queue,

$$\tau =\frac{1}{\overline{W}+1}$$

(10)

Where $\overline{\mathrm{w}}$ is the average number of back-off slots. The probability of direct collision $P_{dc}$ is calculated as follows,

$$P_{dc}={(1-(1-\rho \left)\right(1-\rho \tau )}^{N_{tr}}-1)$$

(11)

Note that $P_b$ represents the probability that a channel is sensed busy when a new packet arrives,

$$P_b=\left( N_{tr}-1\right)\lambda T\left(\frac{1-P_{dc}}{2}\right)$$

(12)

Where T is the complete transmission time of a packet including DIFS period.

Finally, the ${PDR}_{dc}$ for direct collision case is,

$$PDR=1-P_{dc}$$

(13)

As for the hidden terminal case, let $P_{hc}$ represents the probability of a hidden terminal collision,

$$P_{hc}=1-\left(1- P_{dc }\right)P\left(S_1\right) P \left(S_2\right)$$

(14)

Where, S1 denotes the event where none of the hidden terminals transmit, considering the number of hidden terminals is $N_{ph}$ this probability can be expressed as,

$$P\left(S_1\right)=1-N_{ph}\lambda T (1-\frac{P_{dc}}{2})$$

(15)

and $S_2$ denotes the case where a vehicle starts its transmission,

$$\left(S_2\right)=e^{-}\lambda N_{ph}(t_{data}-t_{DIFS})$$

(16)

Where, $t_{data}$ is the transmission time for a packet, and $t_{DIFS}$ is the duration of DIFS period.

Finally, the ${PDR}_{hc}$ for hidden terminal case is expressed as,

$$PDR=1-P_{hc}$$

(17)

To compute the PDR in direct collision and hidden terminal scenarios, we employ Equations 13 and 17, respectively, using the collision parameters outlined in Table 1. The model used in this study, as presented in, examines Safety Applications for vehicles traveling on a multi-lane highway, where the inter-lane distances are negligible compared to the overall network length.

Table 1. MAC model parameters.

Table 1. MAC model parameters.

Average number of back-off slots, W	I6
Transmission range, R	6oo m
DIFS time	64 micro s
Data rate	6 Mbps
BSM rates	IO, 2o and 3o BSM/s
BSS size	i8o Bytes
Vehicle density	z-zoo vehicles/km

The impact of direct collisions on the PDR with varying vehicle densities is illustrated in Figure. 6 With an increase in vehicle density from 2 to 200 vehicles/km, the PDR declines for all three tested BSM rates. However, the impact of direct collisions on the PDR is insignificant for the 10 BSM/s message rate, even at a vehicle density of 200 vehicles/km, with a PDR of 92%. On increasing the BSM rate to 20 BSM/s, the PDR decreases to 72%, and further increasing the rate to 30 BSM/s results in a PDR of only 4%. Such a low PDR would make the Safety Application ineffective at high vehicle densities. The high BSM rates can only be used at lower vehicle densities, e.g., sending 20 BSM/s at vehicle densities below 115 vehicles/km, or 30 BSM/s for 78 vehicles/km.

Figure 6. Impact of direct collisions on PDR (BSM size=180 Byte, data rate=6Mbps, BSM rate= 10, 20, 30 BSM/s).

Figure 6. Impact of direct collisions on PDR (BSM size=180 Byte, data rate=6Mbps, BSM rate= 10, 20, 30 BSM/s).

In this analysis, we examine the effect of hidden terminal collisions on PDR at different vehicle densities, as depicted in Figure 7. The results show that when transmitting at a rate of 10 BSM/s, a PDR of over 90% can only be achieved at a density of 20 vehicles/km. When using higher BSM rates, significant degradation in PDR is observed, which raises concerns about the suitability of the 802.11p MAC layer in dense environments.

Figure 7. Impact of collisions resulting from hidden terminals (BSM size=180 Byte, data rate=6Mbps, BSM rate=10, 20,30 BSM/s).

Figure 7. Impact of collisions resulting from hidden terminals (BSM size=180 Byte, data rate=6Mbps, BSM rate=10, 20,30 BSM/s).

7. Jammer Fail-Safe Mode and Recovery Algorithm

Figure 7. The jammer recovery model Algorithm.

Figure 7. The jammer recovery model Algorithm.

8. Performances Evaluation

Table 8. Field test parameters.

8.1. Indigenous PDR

The experiment aimed to assess the impact of jamming on the Packet Delivery Ratio (PDR) of communication between high and regular vehicles. A jamming detection algorithm was utilized to forecast potential outcomes. The experiment was conducted in an unobstructed area to ensure accurate outcomes. The PDR was measured under normal (non-jamming) communication conditions as the distance between Basic Safety Messages (BSMs) received by the high and regular vehicles was increased.

The findings of the PDR measurements are displayed in Figures 9 and 10, demonstrating that the experimental results align with the predicted evaluations.

Figure 9. Estimated PDR for 3 Mbps.

Figure 9. Estimated PDR for 3 Mbps.

Figure 10. Estimated PDR for 6 Mbps.

Figure 10. Estimated PDR for 6 Mbps.

8.2. The Effect Jamming for PDR

To evaluate the impact of jamming on PDR, a test was conducted with two vehicles (RV followed by HV) driving straight on a two-lane road with a stationary jammer present on the road. During the test, BSM messages were logged by the OBU in the HV for data rates of 3 and 6 Mbps, while being subjected to deceptive jamming at rates of 3, 6, and 12 Mbps. It should be noted that a data rate of 12 Mbps was deemed inappropriate for BSM communication in the presence of jamming in previous studies[31,32]. Figure 11 illustrates the PDR for 3 Mbps BSM communications under different jamming rates in a standard test scenario. As the HV and RV approached the jammer stationed at 600m, the HV could not receive BSMs at a distance of about 375-425m. The impact of the jammer diminished at around 750-800m. The transmission rate of the jammer had only a modest effect on the PDR. It is not recommended to draw conclusions from small variations in these experiments due to the nature of this type of jammer.

Figure 11. PDR at 3 Mbps with deceptive jamming.

Figure 11. PDR at 3 Mbps with deceptive jamming.

The assessment of the impact of jamming on the PDR was carried out by testing two vehicles (RV followed by HV) driving on a straight two-lane road, while a deceptive jammer was located in a stopped car on the side of the road. During the tests, BSM messages were logged by the OBU in the HV for data rates of 3, 6 Mbps while being subjected to deceptive jamming with rates of 3, 6, and 12 Mbps. It should be noted that a data rate of 12 Mbps is considered unsuitable for BSM communication in the presence of jamming. Figure 11 shows the PDR for 3 Mbps BSM communication at different jamming rates for a typical test scenario. As the HV and RV approached the jammer stationed at 600m, the HV was unable to receive BSMs from about 375-425m. The effect of the jammer dropped off at around 750-800m, and the transmission rate of the jammer had only a modest effect on the PDR. However, it should be noted that a sample size of these small variations is needed for further experimentation. The results for the normal test with a data rate of 6 Mbps are shown in Figure 12. Again, the PDR was only modestly influenced by the rate of the deceptive jammer. Interestingly, a peculiar situation was observed during the 3 Mbps jamming test. After the HV was caught in the jamming zone, it was able to receive messages from the RV again at around 475m. This was because a small truck passed the test cars and positioned itself temporarily between the vehicles and the jammer, thereby reducing the effect of the deceptive jammer.

To sum up, the field test results indicated that the transmission rate of the deceptive jammer had a minor impact on the transmissions. However, further tests are necessary to determine how the transmission of different data rates is affected. Although the overall effect of jamming was significant, no clear pattern in the impact of the jamming on different data rates of the vehicles and the jammer was observed. This is in contrast to continuous jamming, which significantly reduces the PDR for higher data rates.

Figure 12. PDR at 6 Mbps on jamming deceptive.

Figure 12. PDR at 6 Mbps on jamming deceptive.

8.3. Jamming Detection Evaluation of Algorithm

The outcomes of the jamming detection algorithm assessment for the 3, 6, and 12 Mbps region test data are shown in Figure 12. The algorithm was successful in detecting jamming as soon as the PDR drop was identified by the consistency test, which was based on distance and PDR. The algorithm did not detect any inconsistencies in distances and GPS, indicating that only one of the two detections mechanisms used in the region test was sufficient. Moreover, PDR inconsistencies were detected.

Figure 13. of Jamming Attack Evaluation Algorithm in Data rate.

Figure 13. of Jamming Attack Evaluation Algorithm in Data rate.

9. Discussion and Results

9.1. Constant Jammer

Consider the scenario illustrated in Figure 14, where an RV is followed by an HV on a single-lane road. Suppose that the RV suddenly brakes due to a hazard, resulting in the dissemination of BSMs carrying braking information to surrounding vehicles. Assuming that the vehicles' speed is 35 mph (15.6 m/s), and the safety distance between them is 3 seconds, the reaction time is 1 second (typical reaction times are around 0.95 seconds). Thus, the HV will have only 2 seconds to receive BSMs about the braking event before it needs to react. We will assume a constant jammer as the source of a malicious attack in this scenario, positioned behind the HV,, we define the unreliability Q(t) of the Safety Application as the inability of the HV to successfully receive at least one BSM before $t_{react}=1 s$, The inability of the HV to receive BSMs from the RV is directly attributed to the jammer's signals overpowering the legitimate communication signals. We will now analyze how BSM rates, power levels, and data rates affect the unreliability Q(t) of the Safety Application. It is important to note that we assume all BSM communication occurs in the safety channel (CH172).

Figure 14. The effect of BSM rates on the reliability of the safety application.

Figure 14. The effect of BSM rates on the reliability of the safety application.

9.2. The Impact of BSM Rats

In order to investigate the effect of BSM rates on the reliability of the safety application, we examined three different BSM rates: 10, 20, and 40 BSM/s, as illustrated in Figure 14. During this experiment, the transmission power was set to P_t=21 dBm, the jammer power was set to P_j=15 dBm, and the data rate was set to R = 6 Mbps. The figure demonstrates how jamming affects Q(t) for different BSM rates. As shown, Q(t) almost reaches 1, indicating complete failure, for the entire time frame leading up to 0.4s before t_react. High BSM rates demonstrate me improvement, but unreliability only decreases when there is almost no time left to react. BSM rates of 10 and 20 BSM/s resulted in unacceptable unreliability of more than 0.2 and 0.45, respectively. At t_react (0 in the figure), only a message rate of 40 BSM/s met the safety application's unreliability requirements with $Q\left(t_{react}\right)=0.04$, implying a safety application reliability of 0.96. However, generally, this is too close to the threshold for reacting. On the other hand, in the context of saving lives using safety applications, this could still be helpful. The above analysis only considers BSM rates and does not take into account the impact of other adjustments, such as transmission power and data rates, which will be discussed next.

Figure 15. The impact of BSM Rates on Q (t).

Figure 15. The impact of BSM Rates on Q (t).

9.2. The effect of Transmission Power.

To examine the impact of transmission power levels on Q(t), we tested three power levels: 21 dBm, 23 dBm, and 25 dBm. For this experiment, we kept the BSM rate fixed at 40 BSM/s, the jammer power $P_j$ was set to 15 dBm, and the data rate was set to R = 6 Mbps. In Figure 16, the impact of different transmission power levels on Q(t) is presented. The experiment was conducted by examining three power levels, namely 21 dBm, 23 dBm, and 25 dBm, while keeping the BSM rate fixed at 10 BSM/s, the jammer power at 15 dBm, and the data rate at 6 Mbps. The results show that, for a transmission power of 21 dBm, the unreliability remains high until around 0.4s prior to $t_{react}$, making it insufficient for the safety application. However, when the transmission power is increased to 23 dBm, Q(t) starts dropping earlier, around 0.9s prior to $t_{react}$, and reaches an acceptable unreliability level around 0.4s before $t_{react}$. Further increasing the transmission power to 25 dBm results in even earlier drop in Q(t), starting around 1.4s prior to $t_{react}$, and reaching acceptable Q(t) about 0.9s before $t_{react}$. The improvement in reliability is attributed to the increase in Signal-to-Jamming Ratio (SJR) as the transmission signals become stronger. This increase allows the safety application to receive at least one BSM before $t_{react}$, contributing to more time for drivers to react. It is worth noting that the chosen transmission power levels are in line with the FCC amendment [33], which states that transmission power levels for public safety operations in CH172 should not exceed 33 dBm EIRP.

Figure 16. The impact of Transmission Power on Q(t).

Figure 16. The impact of Transmission Power on Q(t).

9.2. The effect of Data Rate.

To investigate the effect of data rates on Q(t), two data rates were examined, namely 3 Mbps and 6 Mbps. During the experiment, the power of the jammer, P, was set to 15 dBm, the transmission power was set to P = 21 dBm, and the BSM rate was fixed at the standard 10 BSM/s. Due to their unreliability in the face of constant jamming[34], higher data rates were not considered. Figure 16 illustrates how jamming affects the reliability of the safety application for the two data rates. When using 6 Mbps, Q(t) begins to decline only 0.4s prior to $t_{react}$ and never reaches an acceptable level of unreliability. However, when the lower data rate of 3 Mbps was used, Q(t) starts to decline before 1.1s and meets the application's unreliability requirements before 0.6s from $t_{react}$, providing the driver with additional time to react. The benefit of employing lower data rates stems from the fact that 3 Mbps data rate employs Binary Phase Shift Keying (BPSK) with a coding rate of 1/2, while 6 Mbps uses Quadrature Phase Shift Keying (QPSK) with a coding rate of 1/2, as specified by the ASTM E2213 standard. Higher modulation modes are generally more susceptible to transmission errors.

Figure 17. The impact of Data Rate on Q(t ).

Figure 17. The impact of Data Rate on Q(t ).

9.5. Deceptive Jammer.

In Subsection 9, we examine the recovery concepts for the case of a deceptive jammer. Specifically, we investigate how the BSM rates, transmission power levels, and data rates impact the recovery time. While in the case of a constant jammer, we measured the impact of each parameter on the Safety Application using the unreliability$Q_t$, in the case of a deceptive jammer, we conducted actual field experiments to obtain the results. Therefore, we could no longer calculate the individual $Q_t$which represents the probability that ${BSM}_i$was not received$t_i$. Instead, we used the recovery time, defined as the time required for the HV to resume steady reception after passing the jammer, to measure the impact of BSM rates, transmission power, and data rates. It is important to note that we considered the communication fully recovered only when a steady reception was resumed, even though intermittent reception of BSMs may occur after passing the jammer. The field experiments were conducted on a straight 2-lane road with an average speed of 35 mph (15.6 m/s), with an RV followed by an HV passing a stationary deceptive jammer on the roadside. The moderate speed allowed us to better understand the impact of the tested parameters in the presence of the jammer while not exceeding the speed limit of the test road. Additionally, a third vehicle was included to collect extra data and investigate its impact on the two communicating vehicles when leaving the jammed zone. We focused on the HV's reception of the alert messages during the experiments, as our main concern was the vehicle receiving the messages. Figure 18 shows the position of the test vehicles, with the RV being the first vehicle exposed to the jammer's impact in Figure 18a. The impact of the jammer on the RV is different when the vehicles have passed, as shown in Figure 18b, where the last two vehicles provide a shielding effect on the RV. The figures presenting the results of the field test depict the impact of the jammer on the RV in two scenarios, i.e., moving towards or leaving from the jammer position. Table 3 provides details on the specific parameters utilized in the field tests. However, it is important to acknowledge the challenges involved in conducting these experiments. Conducting field experiments presents several challenges, such as ensuring safety during the tests, controlling testing conditions, and obtaining reliable and accurate data. To address these challenges, the test road was carefully selected, safety measures were implemented, and various data collection methods were employed, such as GPS and video recordings. Furthermore, due to the unpredictable nature of road and traffic conditions, multiple tests were conducted to obtain statistically significant results. The collected data were then analyzed to draw conclusions regarding the impact of BSM rates, transmission power levels, and data rates on recovery time.

Table 3. Field test parameters.

Table 3. Field test parameters.

OBU Model	Arada Systems LocoMate Classic
Vehicle speed	20 m/s
Range test	Two lines for the road
Test range length	1000 m
Jammer position	500 m from starting point
BSM generation	10,20 and 40 BSM/s
Channel	Safety Channel 172
Effective bandwidth	8.3 MHz
Transmitter power	21,23 and 25 dBm
Data rate Transmission	3 and 6 Mbps
Data rate Jammer power	6 Mbps and 18 dBm

Figure 18. The position of the test vehicles prior to and after encountering the jammer.

Figure 18. The position of the test vehicles prior to and after encountering the jammer.

9.6. The Impact of BSM Rates.

To examine how different BSM rates affect recovery time, we conducted field tests using rates of 10, 20, and 40 BSM/s. Results for each rate are displayed in Figures 20, and 21, respectively. The number of BSMs received by the HV was measured over the entire test area (Figure 24) from start to finish, with the time at which the HV passes the jammer indicated by a dashed line. Although passing times varied slightly across trials due to slight speed variations, recovery times were consistent within each test. Figure 19 shows results for a typical experiment using the standard 10 BSM/s rate. When the HV passed the jammer at t = 31s, no BSMs were received because the jamming impact was at its peak and the HV was nearly parallel to the jammer. Steady reception of BSMs resumed only after 12 seconds following the passage of the jammer, with only intermittent reception prior to that time.

Figure 20 illustrates the impact of sending at 20 BSM/s on the recovery time. During this experiment, the RV increased its sending rate to 20 BSM/s, and the HV passed the jammer at t = 31s. As shown in the figure, the HV resumed steady reception at t = 41s, resulting in a recovery time of 9s for this trial.

In the last trial, 40 BSM/s was used, and the HV passed the jammer at t = 29s, as depicted in Figure 21. The HV started receiving BSMs continuously from the RV at t = 39s, leading to a recovery time of 9s.

Figure 19. Reception using 10 BSM/s.

Figure 19. Reception using 10 BSM/s.

Figure 20. Reception using 20 BSM/s.

Figure 20. Reception using 20 BSM/s.

Figure 21. Reception using 40 BSM/s.

Figure 21. Reception using 40 BSM/s.

9.7. The Impact of Transmission Power.

To investigate the impact of transmission power on the recovery time in the presence of a deceptive jammer, three different power levels have been investigated, i.e., 21, 23 and 25 dBm. The impact of these

Studying the impact of different power levels on recovery time in typical test runs is presented in Figures 22, 23, and 23. Figure 22 illustrates a scenario where transmissions were made at a power level of 21 dBm and the corresponding recovery time. At point t = 31s, the HV passed the jammer and communication was completely disrupted due to the deceptive jammer. However, at point t = 41s, a steady reception of BSMs was observed, resulting in a recovery time of 9s. Figure 23 depicts the case of transmission power level of 23 dBm, where the HV passed the jammer at point t = 28s, and steady reception was only resumed after 7s from the point of passing the jammer. Figure 24 considers a transmission power level of 25 dBm. The HV passed the jammer at t = 28s and regained steady reception at t = 34s, resulting in a recovery time of 6s.

Figure 22. Reception using 21 dBm transmissions Power.

Figure 22. Reception using 21 dBm transmissions Power.

Figure 23. Reception using 23 dBm transmissions Power.

Figure 23. Reception using 23 dBm transmissions Power.

Figure 24. Reception using 25 dBm Transmissions Power.

Figure 24. Reception using 25 dBm Transmissions Power.

9.8. The Impact of Data.

To study the effect of data rates on the recovery time, two different rates, namely 3 Mbps and 6 Mbps, were tested, and the results are presented in Figures 24 and 25, respectively. Figure 25 shows an example of the recovery time using a data rate of 3 Mbps, where the HV passed the deceptive jammer at t = 30s and resumed reception of BSMs from the RV at 42s, resulting in a recovery time of 11s. In contrast, Figure 26 depicts the recovery time using a data rate of 6 Mbps, where the HV passed the jammer at t = 31s, and a constant reception of BSMs was regained at t = 44s, accounting for a recovery time of 12s.

However, an abnormal behavior was observed during the field experiments, as spikes in the number of received BSMs were detected. This behavior suggests that the number of received BSMs exceeded the number of transmitted ones. For instance, in Figure 26 at t = 24s, the number of received BSMs was more than 25, while only 10 BSMs were transmitted. This unusual behavior is due to the deceptive jammer preventing the OBU from accessing the media, which causes packets generated by the application layer to accumulate in the OBU's queue for deferred sending. When the OBU eventually gains access to the media, the queued packets are transmitted simultaneously, resulting in the observed spikes. This effect is particularly noticeable at the beginning of the test period when the communication was partially affected by the jammer.

Figure 26. Reception using 3 Mbps data rate.

Figure 26. Reception using 3 Mbps data rate.

Figure 27. Reception using 6 Mbps data rate.

Figure 27. Reception using 6 Mbps data rate.

Upon analyzing the data collected from the field experiments, we noticed abnormal spikes in the number of received BSMs, which exceeded the number of transmitted BSMs. This behavior can be observed in Figure 26, where at t = 24s, the number of received BSMs was more than 25, while the number of transmitted ones was only 10. This abnormal behavior is due to the fact that the deceptive jammer is preventing the OBU from accessing the media. As a result, the OBU queues the packets generated by the application layer for deferred sending. However, since the jamming persists, packets were not sent in time and accumulated over a period. This is especially evident at the beginning of the test period when the communication was partially affected by the jammer. Once the OBU gains temporary access to the media, all queued packets are pushed at once, resulting in the observed spikes. It's worth noting that the queue has a certain capacity, which prevents it from buffering all packets during prolonged inaccessibility to the media.

Table 4 presents the recovery times and distances between HV and RV for the representative cases. However, due to unavoidable differences in distances and environmental conditions, the observed results cannot be generalized.

Table 4. Recovery times for deceptive jammer of the trials presented.

Table 4. Recovery times for deceptive jammer of the trials presented.

Deceptive Jammer (Field Experiment}	BSM /s			Power (dBm)			Data Rate (Mbps)
	10	20	40	21	23	25	3	6
Recovery Time (Seconds)	12.0	9.0	9.0 9.0	9.0	7.0	10.0	11.0	12.0
Distance (meters)	145	115	125	106	68.5	120	125	150

The results obtained from the presented scenarios confirm the predictions made by the mathematical models and intuition. For instance, the increase in recovery time when the power level was increased from 23 dBm to 25 dBm was due to the actual distances observed in the post-analysis of the data. Therefore, the reader should consider the "Distance" column when examining the recovery times. Despite our best efforts to maintain consistent distances between different trials, it was challenging to achieve this without a towrope between vehicles. Nonetheless, due to the unavailability of a facility that could accommodate such a test range length, we were unable to maintain consistent distances.

Also, Table 4 shows the recovery times and distances for the representative cases that were examined. But, due to differences in environmental conditions and distances, these results are not conclusive and cannot be generalized. To provide a fair comparison with similar test conditions, one can simulate different BSM rates in a post-analysis based on a single field trial. Figure 27 shows the comparison of different BSM rates (10, 20, and 40 BSM/s) using the data from the field test with 40 BSM/s. This allows us to understand the impact of BSM rates while maintaining almost the exact same test conditions, such as environmental and physical factors, including distances and antenna positions. However, the conditions using this approach are not entirely identical, as the transmitter queue behavior for 40 BSM/s is unlikely to be the same for rates of 10 and 20 BSM/s. Our results indicate that sending at a rate of 10 BSM/s resulted in a 11s recovery time, while sending at higher rates of 20 and 40 BSM/s resulted in a shorter recovery time of 9s. These results are consistent with our previous observations during the field experiments, as summarized in Table 4. However, we were unable to directly observe the impact of transmission power levels for the deceptive jammer due to variations in inter-vehicle distances during the field experiment. These variations in distances resulted in different levels of SNR, which in turn impacted the reception of BSMs and affected the change in recovery times that we observed during the field test.

Figure 27. Comparing the impact of BSM rates.

Figure 27. Comparing the impact of BSM rates.

Figure 28. The impact of relative distance between vehicles on SNR.

Figure 28. The impact of relative distance between vehicles on SNR.

To better understand the impact of higher transmission powers, we analyzed the relationship between signal-to-noise ratio (SNR) levels and inter-vehicle distances, as shown in Figure 28. By assuming a fixed distance of 100 m between the vehicles, we were able to observe the real impact of using higher transmission power. Comparing the results for 23 dBm and 25 dBm at a distance of 100 m, we found that using a higher transmission power resulted in improved SNR levels. Higher SNR levels translate to lower bit error rates (BER) and overall higher chances of successfully receiving messages. In contrast to the field experiments with the deceptive jammer, where maintaining consistent physical conditions was challenging, we were able to control the physical conditions in this analysis. Therefore, these results provide valuable insights into the potential benefits of using higher transmission powers in similar scenarios.

10. Conclusion

References

1. N. 2015. ] Traffic Safety Facts: Crash Stats, U.S. Department of Transportation, National Highway Traffic Safety Administration, DOT HS 812 219, “No Title.”.
2. “J. B. Kenney, Dedicated Short-Range Communications (DSRC) Standards in the United States, Proceedings of the IEEE, vol. 99, no. 7, pp. 1162-1182, 2011.”. [CrossRef]
3. “No TitleIEEE Standard for Wireless Access in Vehicular Environments - Security Services for Applications and Management Messages, IEEE Std 1609.2TM, 2018.”.
4. Mahmood A. Al-Shareeda; Selvakumar Manickam, “A Systematic Literature Review on Security of Vehicular Ad-Hoc Network (VANET) Based on VEINS Framework,” IEEE 66th Veh. Technol. Conf., vol. 11, pp. 46218–46228, 2023. [CrossRef]
5. S. Blessy, A.M.C., Brindha, “Energy-efficient fuzzy management system using tri-parametric methodology in vanet,” IEEE, vol. 14, 2023. [CrossRef]
6. G. Ayoob, A., Su, G. and Al, “Hierarchical Growing Neural Gas Network (HGNG)-Based Semi cooperative Feature Classifier for IDS in Vehicular Ad Hoc Network (VANET,” J. Sens. Actuator Networks, vol. 7, no. 3, p. 41, 2018. [CrossRef]
7. Jinsong Zhang; Kangfeng Zheng; Dongmei Zhang; Bo Yan, “AATMS: An Anti-Attack Trust Management Scheme in VANET,” IEEE Access, vol. 8, no. 10, 2020. [CrossRef]
8. M. Hadded, P. Muhlethaler, A. Laouiti, R. Zagrouba, and L. A. Saidane, “TDMA-Based MAC Protocols for Vehicular Ad Hoc Networks: A Survey, Qualitative Analysis, and Open Research Issues,” IEEE Commun. Surv. Tutorials, vol. 17, no. 4, pp. 2461–2492, 2015. [CrossRef]
9. G. Lu, Vasukidevi, “A Survey on Security and Key Management in,” IJIRST –International J. Innov. Res. Sci. Technol., vol. 3, no. 08, pp. 107–111, 2017.
10. Y. Hu, A. Perrig, and D. B. Johnson, “Ariadne : A Secure On-Demand Routing Protocol for Ad Hoc Networks,” Wirel. Networks 11, vol. 5, pp. 21–38, 2015. [CrossRef]
11. F. , Xu Hao , J. Hortelano , Sakiz and S. Sen, “Ad Hoc Networks Survey paper A survey of attacks and detection mechanisms on intelligent transportation systems : VANETs and IoV,” Ad Hoc Networks, vol. 61, pp. 33–50, 2017. [CrossRef]
12. H. M. Maier, “Florida State University Libraries Nutritional Status and the Relationship of Dietary and Serum Advanced Glycation End-Products with Inflammation , Oxidative Stress and Healing of Diabetic Foot Ulcers,” 2013.
13. S. I. Sou and O. K. Tonguz, “Enhancing VANET connectivity through roadside units on highways,” IEEE Trans. Veh. Technol., vol. 60, no. 8, pp. 3586–3602, 2017. [CrossRef]
14. E. F. Ahmed Elsmany, M. A. Omar, T. C. Wan, and A. A. Altahir, “EESRA: Energy efficient scalable routing algorithm for wireless sensor networks,” IEEE Access, vol. 7, pp. 96974–96983, 2019. [CrossRef]
15. J. Puñal, Ó., Pereira, C., Aguiar, A., & Gross, “Experimental characterization and modeling of RF jamming attacks on VANETs.,” IEEE Trans. Veh. Technol., vol. 64, no. 2, pp. 524–540, 2015. [CrossRef]
16. Y. Shi, “LTE-V: A Cellular-Assisted V2X Communication Technology,” 2019.
17. F. B. and A. P. A. Studer, E. Shi, “TACKing Together Efficient Authentication, Revocation, and Privacy in VANETs, in Sensor, Mesh and Ad Hoc Communications and Networks,” Access IEEE, pp. 1–9, 2018. [CrossRef]
18. H. L. and M. G. Jie Li, “A Novel Authentication Framework with Conditional Privacy-Preservation and Non-Repudiation for VANETs,” IEEE Trans. Parallel Distrib. Syst., vol. 26, pp. 938–948, 2019. [CrossRef]
19. J. P. and Z. Mammeri, “Authentication and Consensus Overhead in Vehicular Ad Hoc Networks, in Telecommunication systems,” IEEE 66th Veh. Technol. Conf., vol. 52, no. 4, pp. 2699–2712, 2015. [CrossRef]
20. M. G. and Z. C. Z. Cao, J. Kong, U. Lee, “Proof-of-relevance: Filtering False Data via Authentic Consensus in Vehicle Ad-hoc Networks, in IEEE INFOCOM,” Workshops, vol. 20815, pp. 1–6.
21. U. Rajput, F. Abbas, H. Eun, and H. Oh, “A Hybrid Approach for Efficient Privacy Preserving Authentication in VANET,” IEEE Access, vol. 3536, no. c, pp. 1–1, 2017. [CrossRef]
22. P. Donadio, A. Cimmino, and G. Ventre, “Enhanced Intrusion Detection Systems in ad hoc networks using a Grid based agnostic middleware,” Proc. Int. Conf. Pervasive Serv. ICPS 2008 2nd Int. Work. Agent-Oriented Softw., pp. 15–19, 2008.
23. R. G. Engoulou, M. Bellaïche, S. Pierre, and A. Quintero, “VANET security surveys,” Comput. Commun., vol. 44, 2014. [CrossRef]
24. F. Anjum, D. Subhadrabandhu, and S. Sarkar, “Signature based intrusion detection for wireless ad-hoc networks: a comparative study of various routing protocols,” 2003 IEEE 58th Veh. Technol. Conf., vol. 3, 2003.
25. Y. Hu, A. Perrig, and D. B. Johnson, “Rushing Attacks and Defense in Wireless Ad Hoc Network Routing Protocols,” [Computer Commun. Networks, vol. 3.
26. T. O. Address, U. In, and H. Times, “by Maria Isabel Portal Palomo A Thesis Submitted to the Faculty of The College of Engineering and Computer Science in Partial Fulfillment of the Requirements for the Degree of Master of Science Florida Atlantic University Boca Raton , Florida,” no. May, 2013.
27. and C. S. B. Awerbuch, A. Richa, “A Jamming-Resistant MAC Protocol for Single-Hop Wireless Networks, in Proceedings of the twenty-seventh ACM symposium on Principles of distributed computing,” ACM/Springer Mob. Networks Appl., vol. 2015.
28. Y. P. and K. H. R. C.D. Jung, C. Sur, “A Robust Conditional Privacy-Preserving Authentication Protocol in VANET, in Security and Privacy in Mobile Information and Communication Systems,” Springer Berlin Heidelb., p. 35-45, 2014. [CrossRef]
29. A. H. and J. Ben-Othman, “IEEE International Conference on Communications, Dresden,” in Detection of Jamming Attacks in Wireless Ad Hoc Networks Using Error Distribution, 2015, pp. 877–889.
30. K. Sampigethaya, M. Li, L. Huang and R. Poovendran, “Robust Location Privacy Scheme for VANET,” IEEE J. Sel. Areas Commun., vol. 25, no. 8, pp. 569–1589. [CrossRef]
31. Mokdad, L., Ben-Othman, J., & Nguyen, “DJAVAN: Detecting jamming attacks in Vehicle Ad hoc Networks,” Perform. Eval., vol. 87, pp. 47–59, 2015. [CrossRef]
32. M. I. A. S., “Radio Jamming Attacks Against Two Popular Mobile Networks.,” Semin. Netw. Secur., vol. 11( 2), pp. 135–142, 2000.
33. Alasmary, W.; Zhuang, “Public safety application for approaching emergency vehicle alert and accident reporting in VANETs using WAVE,” Ad Hoc Netwowking, vol. 10, p. 90, 2012.
34. W. Chen, R. K. Guha, T. J. Kwon, J. Lee, and Y. Hsu, “A survey and challenges in routing and data dissemination in vehicular ad hoc networks,” Wirel. Commun. Mob. Comput., vol. 15, no. October 2009, pp. 787–795, 2011.