Balancing Awareness and Congestion in Vehicular Networks Using Variable Transmission Power

Vehicular ad Hoc networks (VANETs) support a variety of applications ranging from critical safety applications to “infotainment” or “comfort” applications. In North America, 75 MHz of the spectrum in the 5.9 GHz band has been allocated for vehicular communication. Safety applications rely on event-driven “alert” messages as well as the periodic broadcast of Basic Safety Messages (BSMs) containing critical information, e.g., position, speed, and heading from participating vehicles. The limited channel capacity and high message rates needed to ensure an adequate level of awareness make the reliable delivery of BSMs a challenging problem for VANETs. In this paper, we propose a decentralized congestion control algorithm that uses variable transmission power levels to reduce the channel busy ratio while maintaining a high level of awareness for nearby vehicles. The simulation results indicate that the proposed approach is able to achieve a suitable balance between awareness and bandwidth usage.


Introduction
The number of motor vehicle deaths in the United States, which includes all types of motor vehicles, e.g., passenger cars, trucks, and buses, was almost 40,000 in 2019 [1,2]. Besides driver attitudes and behaviors, vehicle safety technology is a key factor to make car travel safer. VANETs [3] have been widely perceived by government, car manufacturing industries, and academia as a valid measurement to achieve safety and efficiency in the motorway. In VANET, vehicles communicate directly with each other, as well as with infrastructure nodes to create a connected network that can facilitate the dissemination of the necessary safety-related information. Vehicles equipped with communication devices known as On Board Units (OBUs) can connect with other vehicles and RoadSide Units (RSUs) to implement various safety applications.
In North America, 75 MHz of the spectrum in the 5. 9 GHz band have been allocated using Dedicated Short-Range Communications/Wireless Access for Vehicular Environment (DSRC/WAVE) [4,5] for Vehicle-to-Vehicle (V2V) communication, and another 30 MHz of the spectrum is proposed for Cellular Vehicle-to-Everything (C-V2X) [6]. Safety applications rely on up-to-date information on surrounding vehicles, including their position, speed, acceleration, heading, etc. Therefore, each vehicle periodically broadcasts its status, containing such information, in the form of BSMs [7] (also referred to as Cooperative Awareness Messages (CAMs) [8] in Europe). The information contained in these packets are used as input to safety-critical applications, e.g., collision avoidance, lane change warnings, and hard braking warnings. The availability and freshness of the BSM data are essential for the proper operation of the different algorithms designed to improve vehicle safety.
In order to achieve a suitable level of awareness, the target transmit rate for the BSM packets is set as 10 Hz with a transmit radius of 300 m [9]. The total allocated bandwidth for BSM transmission from all vehicles is expected to be fully utilized, with 200 vehicles sending their BSMs at the standard transmission rate [10]. This is for an ideal case that assumes that there are no other types of transmissions and that all BSM packets are perfectly scheduled back-to-back, without any (packet) collisions. The IEEE 802.11p Carrier Sense Multiple Access (CSMA) protocol that is used for V2V communication will actually be able to support far fewer vehicles. The relatively high level of channel congestion [11], even at moderate vehicle densities, means that appropriate congestion control mechanisms are needed that will lead to efficient use of available channel bandwidth and allow a timely delivery of BSMs, while limiting channel congestion as the number of vehicles increases.
The limited channel capacity and high message rates (to ensure an adequate level of awareness) make the reliable delivery of BSMs a challenging problem for VANET, and a number of different congestion control approaches have been proposed for vehicular communication [12]. Message transmission rate used in [10,13,14] and power control used in [15][16][17][18][19] (or a combination of both used in [20][21][22][23]) are the common parameters that are controlled to mitigate channel congestion, although other parameters, e.g., data rate, carrier sense threshold, have also been suggested [24,25]. Both message rate and power control reduce awareness of the neighboring vehicles either by increasing Inter-Packet Delay (IPD) for BSMs received from a vehicle or by limiting the distance at which such packets can be received correctly. In this paper, we propose a new congestion control algorithm that balances the need for high vehicle awareness with maintaining a low channel congestion level. In this approach, each vehicle uses a variable transmission power for each BSM so that nearby vehicles receive more BSMs and more distant vehicles receive fewer BSMs from the same vehicle.
The remainder of the paper is organized as follows. In Section 2, we review existing approaches for congestion control in VANET, and we present our proposed algorithms in Section 3. In Section 4, we discuss and analyze our simulation results, and finally, we present our conclusions and some directions for future work in Section 5.

Related Work Existing Decentralized Congestion Control Approaches
In order to make distant vehicles aware of the Ego Vehicle (EV), i.e., the particular vehicle that sends the BSMs, the transmission power should be increased. Similarly, to ensure that all vehicles have up-to-date information on the EV, it is desirable to increase the number of BSMs sent per second. However, when BSMs are transmitted with high rate and high power, the channel congestion increases quickly as the number of vehicles increases. Therefore, existing congestion control techniques in VANET typically reduce the message rate or transmission power, or a combination of both. In paper [26], the authors propose a comparative review of congestion control mechanisms. Papers [27,28] introduce theory and methods about general traffic information collection and diffusion from the Internet of Vehicle (IoV) point of view.
The nominal message rate for BSMs is 10 packets per second. Rate-based congestion control algorithms attempt to reduce this to a predetermined fixed rate [29] or a new rate that is calculated based on an input parameter which describes the current congestion level [30,31]. One of the most well-known rate-based congestion control approaches is the LInear MEssage Rate Integration Control (LIMERIC) algorithm [10]. The main objective in LIMERIC is to share the available channel bandwidth fairly among all vehicles by adapting the rate of the BSM transmission dynamically in each iteration. LIMERIC is able to effectively mitigate congestion [30], but this is achieved at the cost of increased IPD from neighboring vehicles. In [13], the authors present a new algorithm called Error Model Based Adaptive Rate Control (EMBARC), which updates LIMERIC by also considering vehicle position error estimation. Another rate-based approach, BSM RAte control over IEEE802.11p VEhicular networks (BRAEVE) [14], uses the estimated number of vehicles rather than channel utilization as the input parameter. BRAEVE is shown to provide smoother convergence and result in lower packet error ratio, IPD and tracking error compared to the other algorithms.
In addition to the message rate, another common parameter that can be adjusted to achieve congestion control is the transmission power [32]. A packet sent with a higher power is able to travel a greater distance (and potentially reach more vehicles) compared to one sent with a lower power. This means we can control the number of vehicles that a BSM can reach by adjusting the transmission power. When selecting a suitable transmission power, it is usually calculated based on the measured or estimated level of channel load. Another approach, proposed in [15], is to select the transmission power randomly from a known probability distribution.
One of the most cited power control algorithms is Distributed Fair Transmit Power Adjustment for VANETs (D-FPAV) [16], which proposes a fully distributed, asynchronous and localized approach by setting the node transmission power based on the prediction of application-layer traffic and the observed number of vehicles in the surrounding area. In [33], the authors propose a distributed congestion control algorithm based on noncooperative game theory, which has better performance in terms of fairness and bandwidth usage. In [17], an adaptive beacon transmission power algorithm based on vehicle position prediction error is proposed, which increases the beacon transmission power of vehicles with large vehicle position prediction errors and reduces the transmission power of vehicles with small errors. Other power control based algorithms include Segment-based Power Adjustment for Vehicular environments (SPAV) [18], Speed Based Adaptive Power Control (SBAPC) [34], Adaptive transmit power Cooperative Congestion Control (AC3) [19], etc. Paper [12] gives a comprehensive review for the details about these power control based algorithms.
Existing VANET congestion control approaches generally attempt to mitigate congestion by calculating a message rate or power level based on some input parameters, e.g., Channel Busy Ratio (CBR) and vehicle density. In such approaches, all neighboring vehicles are treated in the same way. In other words, all vehicles within the transmission range will receive the same number of BSMs from the EV. However, in practice, the awareness of vehicles that are nearby should be more important than those that are farther away. This is because interactions with nearby vehicles are likely to occur more quickly, leading to a smaller margin for error. Unlike existing approaches, the congestion control algorithm proposed in this paper balances the awareness of near and distant vehicles, so that more packets are received from nearby vehicles and fewer from vehicles that are farther away. A preliminary version of this work, with some initial results was presented in [35]. In this paper, we have updated the algorithm to separate the power control and rate control components so that each part can be adjusted independently. We have also used a random rate control algorithm in Section 3.2.1, which does not require coordination among different vehicles, and added an illustrative example and comprehensive simulation results, including a range of performance metrics, e.g., average CBR, number of received BSMs, and IPD, that were not considered previously.

Algorithm Design Methodology
From the safety point of view, vehicles that are closer to the EV have more impact on its safety. This is because the time to reach a nearby vehicle is much shorter compared to distant vehicles, leading to a smaller margin for error. Hence, delivering safety messages to the near vehicles in a timely fashion is of critical importance. The actual distance that BSMs from a vehicle can reach, i.e., the transmission range (Tx range) of the BSM, depends on the corresponding transmission power (Tx power). A receiving vehicle within the transmission range of the EV should be able to receive the BSM correctly (assuming there is no channel congestion or other interference). As the transmission power increases, the transmission range also increases. In addition to the Tx power, other factors, e.g., the presence of obstacles and the number of surrounding vehicles, can also affect the range. Table 1 shows how the number of received BSMs between two vehicles varies with the distance between the vehicles.
To generate the above data, we considered two stationary vehicles separated by different distances, e.g., 50 m and 100 m, sending periodic BSMs to each other. We have assumed a simple free space path loss model to approximate the transmission range, which does not consider obstacles. With Tx power = 20 mW, we observe that most BSMs can be delivered within 500 m and almost all BSMs are lost when the two vehicles are separated by 600 m or more. With 5 mW, the Tx range is much shorter due to the lower Tx power. In this case, most BSMs can be received if the vehicle separation is 250 m or less, and beyond that, more and more BSMs will be lost. For 5 mW power, there is actually a gradual drop-off for received BSMs between 250 m and 350 m, similar to what occurs between 450 m and 650 m for 20 mW. However, this is not apparent in Table 1, as the tested distances between vehicles were all multiples of 50 m. A smaller vehicle spacing would allow this effect to be seen more clearly. The values shown in Table 1 represent an ideal case, where there are no other vehicles on the road segment. If we have more vehicles, i.e., a high vehicle density, the valid Tx range will be further reduced due to the interference between the vehicles.  50  5982  5980  2  5980  2  100  5982  5980  2  5980  2  150  5982  5980  2  5980  2  200  5982  5980  2  5980  2  250  5982  5980  2  5969  13  300  5982  5980  2  2872  3110  350  5982  5980  2  0  5982  400  5982  5980  2  0  5982  450  5982  5978  4  0  5982  500  5982  5957  25  0  5982  550  5982  5481  501  0  5982  600  5982  1728  4254  0  5982  650  5982  0  5982  0  5982  700  5982  0  5982  0  5982 In summary, a high Tx power is able to deliver BSMs to distant vehicles, but can also lead to higher interference and channel congestion. The key strategy of this paper is to prioritize BSM delivery to nearby vehicles, which can be achieved using a lower Tx power level, instead of persistently transmitting at the maximum 20 mW Tx power. It is still desirable to maintain some level of awareness for distant vehicles as well, but the frequency of updates can be lower than needed for the closest vehicles. Therefore, by varying the Tx power, it is possible to maintain high awareness for nearby vehicles, while still allowing some BSMs to reach distant vehicles, by using a high Tx power for selected BSMs. With this strategy, the awareness of EVs will remain high as needed for nearby vehicles and the congestion of the whole channel will be mitigated at the time because fewer BSMs are sent with a high Tx power.

Proposed Algorithm
In this section, we discuss our proposed Balancing Awareness and Congestion with Variable Tx Power (BACVT) algorithm for VANET congestion control. With this algorithm, vehicles at close range see all packets at their full rate and those farther away see only those packets sent with a higher transmission power. The algorithm can be implemented assuming a constant message transmission rate (the standard rate is 10 BSMs per second) or in conjunction with a message rate control algorithm, e.g., LIMERIC. Algorithm 1, as shown below, outlines the BACVT algorithm when using the standard BSM message rate. Send packet with Tx c ;

end
During the initialization phase, we first define two transmission ranges d n and d f as the near and far ranges (in m), respectively, with respect to the EV. The corresponding Tx powers needed in order to achieve these ranges are Tx n and Tx f , respectively, where Tx n < Tx f . In practice, the required Tx powers depend not only on the target distance, but additional factors, e.g., atmospheric conditions, obstacles and vehicle density. We note that the proposed algorithms simply assume that suitable values of Tx n and Tx f have been calculated separately to be given as input, and we are not proposing new models to calculate such values. We expect that a vehicle v at a distance d from the EV will receive: The value of TotalSend is set to the number of BSMs to send in each cycle in step 3. Unless otherwise specified, we assume that the duration of each cycle is 1 s. In step 3, we define two transmission rates, R n and R f , as the near and far distance reception rates, respectively, i.e., the number of BSMs received per second from the EV by the near and far vehicles, respectively. Clearly, R n > R f , as nearby vehicles will receive all BSMs, those sent with low power as well as those sent with high power. The parameters R n and R f are used by the algorithm to determine which packets are sent with a higher power level in order to maintain the two perceived reception rates for the remote vehicles. These values depend on the current level of congestion in the network, as well as the awareness level needed for the specific applications. For our simulations, we have specified fixed values for R n and R f , where all vehicles use the same values for R n and R f . However, this is not strictly required and different vehicles can use different values, based on their current needs. In step 4, the value of N low is calculated using Equation (1). N low determines the number of low-power BSMs that should be sent between two high-power BSMs for each cycle. This will allow the high and low power BSMs to be evenly distributed over each cycle. An example of how N low is calculated based on Equation (1) is given in Section 3.3.
Once the value of N low is determined, the algorithm initializes the total number of packets sent so far (PktNum) and the number of Consecutive Low-Power Packets (CLPP) to 0 in steps 5 and 6, respectively. These two values maintain a counter of the total number of packets and the number of consecutive low-power packets that have been sent since the last high power transmission. The first BSM of each cycle is sent using the high Tx power, i.e., Tx f . Then, N low packets are transmitted with the low Tx power, i.e., Tx n , and this pattern repeats until the end of the cycle is reached, i.e., TotalSend number of packets have been sent in the current cycle.
After setting the initial parameters, steps 7-18 are repeated for each BSM to determine the Tx power for the individual BSM. If it is the first BSM in the cycle, or N low consecutive low-power BSMs have been sent, then the Tx power is set to high (step 9), CLPP is reset to 0 and the packet count is updated (step 10 and step 11). Otherwise, the Tx power is set to low (step 13) and both the CLPP values and packet count are incremented (step 14 and step 15). Finally, in step 17, the BSM is broadcast by the EV with the appropriate Tx power. The time complexity of Algorithm 1 is O(n), where n is the number of BSMs sent by a vehicle.

Hybrid Approach
The BACVT approach described in Algorithm 1 assumes that each vehicle will use a constant message Tx rate to send the BSMs (the default is 10 BSMs per second). However, the proposed approach can also be used in conjunction with other rate control algorithms to further reduce congestion, if power adaptation alone is not sufficient. In this case, the initial steps (steps 1-6) must be carried out for each cycle, rather than just once at the beginning. The modified algorithm (BACVT-H), which is a "hybrid" of rate control and power control approach, is very similar to the original algorithm, and is shown in Algorithm 2. The main difference is in step 2, where the value of TotalSend is determined at the beginning of each cycle by a suitable rate control algorithm, e.g., LIMERIC, BRAEVE or any other approach. So, the value of TotalSend is not a fixed quantity (as in Algorithm 1), but will vary based on the current CBR and the rate control algorithm that is being used. We have used TotalSend values between 5 and 10 for the message rate control. Therefore, the expected values for the beacon intervals will vary between 0.1 s and 0.2 s. The remaining parameters are set as before (steps 3-6). Finally, the process for determining the Tx power of each BSM (steps 7-18) is similar to Algorithm 1, except that these steps are repeated only for the current cycle and not continued for multiple cycles. At the end of the cycle, the message rate control algorithm is applied to determine the new value of TotalSend, and the initial parameters are recalculated before BSM transmission starts. Since the only difference is the value of TotalSend, Algorithm 2 has the same time complexity with Algorithm 1, i.e., O(n). Figure 1 shows the pattern of high-power and low-power BSM transmissions for a vehicle, using the following parameters:

Illustrative Example
Then, using Equation (1), we get N low = 4. As shown in the figure, the first packet in the cycle (PktNum = 0) is always transmitted with high Tx power. This is followed by N low = 4 low-power packets and another high-power packet. This pattern continues until the last packet in the cycle (PktNum = 9) and the next packet (PktNum = 10) starts a new cycle.

Simulation Parameters
In this section, we evaluate the performance of the proposed congestion control algorithms (BACVT and BACVT-H) in terms of various metrics. To simulate the proposed algorithms, we use a framework of Vehicles in network simulation (Veins) [36], which contains a basic implementation of the IEEE 802.11p and IEEE 1609 protocols in order to facilitate the testing of V2V networks. Veins connects a widely used network simulation tool, Objective Modular Network Testbed in C++ (OMNeT++) [37], and the traffic mobility simulator Simulation of Urban MObility (SUMO) [38].
In our simulations, the BSM packet size is set to 512 Bytes and the default beacon interval, i.e., 0.1 s, is used for all approaches except BACVT. In other words, each vehicle transmits its BSMs at the rate of 10 BSMs per second. For the BACVT-H algorithm, the beacon rate is randomly selected between 5 and 10 BSMs per second, which means the beacon interval will be between 0.1 s and 0.2 s. The transmission power for the far and near range used in BACVT are set to Tx f = 20 mW and Tx n = 5 mW, respectively. We note that it is possible to use the proposed algorithms with different values of Tx f and Tx n . We have selected 20 mW for high power as it is the FCC standard for wireless transmitters and maximum power that can be used without registering [5]; we selected 5 mW for the low-power value as it is still able to achieve a reasonable transmission range of around 300 m. The data rate for BSM transmissions is set to 6 Mbps, which is the most commonly used value [16]. We use two straight 4-lane highway (2 lanes in each direction) traffic models with two kinds of vehicle densities: 200 vehicles (high density) and 100 vehicles (low density). All vehicles, regardless of direction, share the same communication channel. In order to simulate a steady-state traffic scenario and avoid the impact of vehicles entering and leaving the simulation, we assume that all the vehicles are stationary. A summary of the configuration parameters is given in Table 2. The following algorithms are used to compare the performance. Each algorithm will be evaluated with low and high density traffic scenarios. The low density traffic scenario has 100 vehicles (25 in each lane), with vehicles in the same lane spaced 160 m apart, whereas the high density traffic scenario has 200 vehicles (50 in each lane), with vehicles in the same lane spaced 80 m apart.
BACVT-H (A hybrid approach combining BACVT with random rate control).

Average Channel Busy Ratio
CBR is defined as the ratio between the time the channel is sensed as busy and the total observation time. CBR is an effective indicator of the level of channel load. A higher CBR value typically indicates a higher channel load and vice versa. In order to observe the change of the channel's CBR throughout the simulation process, each vehicle will calculate the current CBR based on the status of the channel it detects before it sends each BSM. Figure 2 shows the individual CBR values observed by each vehicle over the simulation interval for the 10 Hz 20 mW case. Clearly, it is difficult to observe the overall variations in channel CBR from this plot. So, we use a new metric called "Average CBR", which calculates the average value of the CBR from all vehicles in 5 s intervals until the end of the simulation time.  Figures 3 and 4 show the Average CBR values obtained using the different approaches for the low density and high density traffic models, respectively. As expected, 10 Hz 20 mW has the highest CBR, and both BACVT8_2 and BACVT5_5 can reduce the CBR. The 10 Hz 5 mW algorithm has an even lower CBR because the low Tx power results in a significantly shorter transmission range. The BACVT-H approach has the best performance in terms of CBR values and is able to reduce congestion by 25% compared to 10 Hz 20 mW and 20% compared to 10 Hz 5 mW. This is because the total number of BSMs sent per second in this approach is less than all the other algorithms.

Received BSMs over Different Distances
When a vehicle can receive more BSMs, it has more awareness of the other vehicles, which is an important performance metric in vehicular communication. However, the awareness of the near neighbors (e.g., vehicles within 300 m of the EV) is more important than the awareness of distant vehicles (e.g., vehicles more than 1 km away). The target of the proposed algorithm is to improve the awareness of the near neighbors while still maintaining an adequate level of awareness of distant vehicles. Tables 3 and 4 show the number of received BSMs of the different approaches for low and high vehicle density scenarios, respectively. The proposed BACVT8_2 and BACVT5_5 approaches show improved reception rates for low distances, while maintaining some awareness (at lower levels) for higher distances. On the other hand, although the 10 Hz 5 mW approach shows higher reception rates for low distances, the awareness drops to 0 for distances greater than 300 m. The BACVT-H approach maintains a moderate level of awareness for different distances. The above results indicate that with the proposed BACVT algorithm, it is possible to achieve the desired level of awareness for near and far vehicles by appropriately tuning the relative number of high-power and low-power transmissions (i.e., R n and R f values) in each cycle. In this context, we note that 10 Hz 20 mW and 10 Hz 5 mW can be viewed as special cases of BACVT, corresponding to BACVT0_10 and BACVT10_0, respectively.

Beacon Error Rate
The Beacon Error Rate (BER) is the proportion of all BSMs sent throughout the simulation that are lost (i.e., not received successfully by a vehicle). If L is the total number of lost BSMs in the simulation and S is the total number of sent BSMs in the simulation, then BER = L/S. Figure 5 shows the BER values obtained using the different approaches for low density and high density traffic scenarios. All the BACVT algorithms (without rate control) have a lower BER compared to 10 Hz 20 mW, as they will have a lower chance of packet collision. As expected, 10 Hz 5 mW has the lowest BER due to low power transmissions, which means the chance of packet collisions is reduced further. The BER value of BACVT-H is slightly anomalous, since the number of sent BSMs in this case is lower than all the other approaches. So, even though the number of lost packets is smaller compared to most of the other approaches, as shown in Figure 6, the calculated BER value is actually higher.

Inter-Packet Delay
We consider V as the set of all vehicles in the simulation. The IPD for a vehicle pair (s, r), where s, r ∈ V, s = r, is the average delay between successive BSMs from s received by r. Let M s,r be the set of all successfully received BSMs for vehicle pair (s, r) and t s,r,i be the time when the i th BSM i ∈ M s,r from s is received at r. Then the average IPD over all vehicles in the simulation is: (a) BER in low vehicle density (b) BER in high vehicle density To calculate the average IPD with respect to vehicles within a specific range d (d 1 < d ≤ d 2 ) of a vehicle u ∈ V, we define the set V u,d 1 ,d 2 as the set of all vehicles that are within a distance d from u, where d > d 1 and d ≤ d 2 . Then, we update Equation (2) as follows: From Table 1, we see that the number of received BSMs decreases significantly between 250 m and 350 m for 5 mW transmission power. Our simulation results, as shown in Figures 7 and 8, show that all the algorithms (except for BACVT-H) have very similar IPD performance for messages from vehicles within a range of 320 m. After 320 m, BACVT8_2 has the highest delay because it sends more BSMs with low Tx power. However, when we adjust the ratio of high Tx power and low Tx power with BACVT5_5, the value of IPD is improved. As expected, 10 Hz 20 mW has the lowest value of IPD for larger distances, because all the BSMs are sent with 20 mW. Finally, we note that although 10 Hz 5 mW performed well in terms of BER, it has the poorest performance when considering IPD for vehicles beyond 320 m, as no packets are received from such vehicles.  It is interesting to note that for vehicles within 320 m, even with a lower transmission rate, BACVT-H is able to achieve an IPD value that is comparable to (but still slightly exceeding) 10 Hz 20 mW. This is shown in Figures 9 and 10, where BACVT-H has a a slightly higher IPD compared to all the other approaches, which have almost identical performance.

Conclusions
In VANET, the available 10 MHz channel can quickly get congested, and the allocated bandwidth is not sufficient to guarantee the delivery of safety messages without latency and error when vehicle density increases. Reducing the transmission rate or the transmission power can indeed mitigate the congestion. However, this is achieved at the cost of reducing the awareness of surrounding vehicles. In this paper, we proposed a novel congestion control approach, where the goal is to effectively balance the channel congestion with the need for vehicle awareness. The main idea is that awareness of nearby neighbors should be prioritized compared to distant vehicles. We first proposed a variable power control method (BACVT) that uses different power levels for transmitting different BSMs. The simulation results show that BACVT achieves a better balance between awareness and congestion compared to traditional methods. To further mitigate the congestion, we also proposed a hybrid approach BACVT-H, which combines rate control with power control to limit the number of the total BSMs. The results demonstrate that BACVT-H indeed has the lowest CBR, while still keeping a good awareness (IPD) performance. The balance of awareness and congestion is always a challenging topic for safety communication in VANET.
In our future work, we will perform more detailed studies on this topic by considering a target-oriented active hybrid congestion control method based on dynamic prediction of vehicle density. Our goal is to make the vehicles predict the next possible near and distant traffic status based on the current status and adjust the transmission parameters to meet some metrics of congestion and awareness. We also plan to develop accurate propagation models that calculate the transmission powers and transmission rates, based on the target transmission range, vehicle density, obstacles, weather conditions, etc.