Congestion Control in Autonomous Resource Selection of Cellular-V2X

Intelligent transportation systems have recently become a promising technology for future industry to provide safe, green, and automated driving. In this regard, the Third Generation Partnership Project (3GPP) has proposed to utilize cellular communication to enable direct communication between vehicles with and without cellular infrastructure assistance. 3GPP has introduced the Sensing-Based Semi-Persistent Scheduling (S-SPS) technique when coverage of the cellular system is absent. S-SPS faces resource collision and performance degradation problems when channel load increases in the network. This paper suggests a decentralized congestion control and transmission power control mechanism (TPC-DCC) with an adaptive threshold for the received signal as a combination method to decrease channel load in the network. Furthermore, this work introduces a novel channel load adjustment. The new adjusting algorithm is based on a constant difference between the upper and lower boundaries of channel load at each level to handle channel overload and provide more flexibility using DCC mechanisms. The interactions of the proposed algorithm with S-SPS and the Extended-Estimation Reservation Resource Allocation (E-ERRA) algorithms that the authors previously proposed are investigated. The results indicate that system performance can be substantially improved when the transmission power and reception threshold are adaptively adjusted to the proposed channel load adjustment. The results of E-ERRA with the proposed channel load adjustment method show promising results compared to S-SPS.

Message (CAM) has been suggested by ETSI [6] as a periodic message broadcast to surrounding vehicles. This message contains information about speed, movement direction, type, acceleration, and deceleration. IEEE defines it as a Basic Safety Message (BSM) or a Beacon [7], [8].
The CAM message consists of two parts in adjacent or nonadjacent configuration. In two resource blocks, the first part carries the SideLink Control Information (SCI) to help the receiver decode the second part. The second part contains Transport Blocks (TB) within a variable number of resource blocks. Furthermore, SCI and TB are transmitted using the Physical Side Link Control Channel (PSCCH) and the Physical Side Link Shared Channel (PSSCH), respectively [9].
3GPP proposed Sensing-based Semi-Persistent Scheduling (S-SPS) as a Mode 4 scheduling method with the use of the Single Carrier Frequency Division Multiple Access (SC-FDMA) technique. The authors of this article have previously suggested Estimation and Reservation Resource Allocation (ERRA) and Extended-ERRA algorithms as alternative scheduling methods to reduce resource collisions and complexity in the selection and reselection of radio resources [10], [11].
An increased channel load in the given band in the broadcasting range of a vehicle can cause resource collisions and hidden nodes, especially when the number of broadcasters is greater than the limited number of resources. The restricted number of resources depends on several factors such as the data rate and the amount of data that are related to the number of resource blocks in each subchannel [8], the transmission rate, and the coverage area resulting from the transmission power.
In [12] and [13], Decentralized Congestion Control (DCC) mechanisms are proposed for VANET communication to deal with the increasing channel load issue by adapting the parameters of the physical layer throughout such as transmit power, transmit rate, and data rate. However, the technical standard did not clearly explain the mechanism for selecting the transmit resources for packets when the channel load is high and changes rapidly. Generally, little attention is paid in the literature to investigating interactions between DCC and scheduling methods for C-V2X communication.
The DCC mechanisms consist of three parameter adjusting mechanisms to regulate the channel load as mentioned below. These mechanisms can adjust the transmission power, data rate, or transmission rate based on the Channel Busy Ratio (CBR). Furthermore, the DCC mechanism is characterized by three types of states: relaxed, active, and restrictive. In turn, the active state contains one or more active levels that differ in physical parameters. This work proposes an adjusting algorithm to support several levels of CBR. A fixed relative difference between the upper and lower boundaries of CBR levels provides more flexibility and excellent immunity against high channel overload. The levels of transmission power adaptation using the transmission power mechanism are synchronized with the signal reception sensitivity thresholds to adjust the channel load for the LTE-V2X Mode 4 vehicular communication system. However, the CBR could be understood as the channel utilization. Generally, channel utilization in V2X communication is defined as the ratio of the number of busy subchannels to the total number of subchannels during the broadcast period. The maximum value of channel utilization is usually called channel capacity [14] and [15]. Therefore, 100% channel utilization indicates that there is no free subchannel to transmit data.
This work investigates the interaction of the proposed method with the S-SPS and E-ERRA radio resource scheduling methods in terms of the packet reception ratio and the packet collision ratio. To provide more realistic environments, the simulation has considered two highway scenarios and MCS orders in fixed and varying vehicle densities.

A. RELATED WORKS
A number of research groups have paid attention to the DCC mechanisms applied in C-V2X. One of the first investigations on the DCC mechanisms that enable C-V2X Mode 4 was carried out by Mansouri et al. [16].
Based on the table that contains the fixed number of channel load adjustments that 3GPP proposes, Mansouri investigated the mechanism of dropped packets when the S-SPS scheduling method is applied for LTE-V2X with a fixed number of vehicles.
Furthermore, Mansouri highlighted the S-SPS problem of resource selection when the channel load increases and proposed a packet drop technique to reduce the channel busy ratio. In [17], the authors analyzed the efficiency of the packet drop technique to improve QoS in the application layer and tested the proposed method for several vehicle speeds. Consequently, the authors found that dropping packets as congestion control harms performance at the application layer.
In [19], interactions of DCC mechanisms with the LTE-V2X Mode 4 resource allocation algorithm were investigated by changing the resource reservation interval (RRI) within S-SPS with the measured CBR or channel occupancy. Bazzi et al. in [21] recommended further investigations when increasing the frequency of awareness message transmissions without influencing vehicle awareness. Adaptive resource (re)selection with a resource collision detection algorithm was proposed in [22]. In [23], the authors of this article introduced a spectrum partitioning technique to avoid the overlap area of the broadcast in C-V2X Mode 3.
The multi-hop clustering algorithm was investigated using a hybrid architecture with infrastructure support [29]. An analytical model for autonomous Cellular-V2X Mode 4 was investigated using the Veins software platform [30] to compare performance at different vehicle densities. VOLUME 11, 2023 B. PAPER ORGANIZATION The rest of this paper is organized as follows. The assumptions and general background of C-V2X are presented in Section II. The proposed algorithm applied in S-SPS and E-ERRA is shown in Section III. Section IV presents the system model and simulation scenarios. The system performance and simulation results are described in Section V. Section VI concludes the paper.

A. CONGESTION CONTROL
In [4], the application of congestion control in LTE-V2X Mode 4 has been specified. The standard does not describe a precise algorithm for controlling channel congestion but presents corresponding metrics and possible mechanisms to decrease the channel load that leads to resource congestion. In each packet (re)transmission, the vehicle has to measure the SL Channel Busy Ratio (CBR) and SL Channel occupancy Ratio (CR). Measurements should be performed in the subframe τ -4 when the transmission of the packet will be broadcast on the subframe τ . CBR indicates the level of channel load, which can be defined as the number of subchannels in the last 100 ms (100 subframes) that feature an average of the Received Signal Strength Indicator (RSSI) higher than a pre-established threshold. On the other hand, CR quantifies the channel occupancy caused by vehicle broadcast packets. It defines the number of subchannels that the vehicle utilizes to transmit during an interval of 1000 ms (1000 subframes). Each vehicle can divide this time period into two intervals, the past and future intervals, and can decide how many past and future subframes for CR are used for measurement. However, the number of past subframes should be equal to or greater than 500. Moreover, only the subframes already used by the broadcasting vehicles can be considered to measure CR. The standard proposed that up to 16 CBR levels can be defined. Vehicles cannot cross CR limit , which increases as the CBR decreases at each level. The value of CR limit is a function of the priority of the TB and the absolute speed of the vehicle. Each vehicle needs to measure CBR and quantifies its CR before the (re)transmission of the packets. If the measured CR is higher than the configured CR limit established for the current level of CBR, the vehicle must reduce this CR to the value below CR limit by adjusting the transmission parameters. However, to achieve this reduction, the standard indicates several potential mechanisms [12]: • Transmit power control (TPC): In TPC, the transmission power changes to adjust to the current channel load. For example, a vehicle decreases its transmission power when a high channel load is observed, reducing the interference range and diminishing the CBR.
• Transmit rate control (TRC): This mechanism works by adjusting the time between two consecutive transmission packets sent by a vehicle.
• Transmit data rate control (TDC): This mechanism can be used by a vehicle that can transfer data in several rate options. During high channel load time periods and depending on the application, a car can adjust the data rate of the SL transmission to be appropriate for the current channel condition. It is essential to mention that the standard did not specify the ranges of CBR levels and the weights of CR limit for each level. However, 3GPP introduced a working document considering the ranges of CBR levels. However, our work proposes a new algorithm to assign CBR to each level, compared to the algorithms shown in [16] and [19].

B. C-V2X COMMUNICATION ASSUMPTIONS
As mentioned above, SC-FDMA is applied to periodically broadcast CAM messages in the 10 MHz channel bandwidth. Each subframe consists of two slots (0.5 ms each). In turn, each subchannel consists of a group of Resource Blocks (RB), which are the smallest units allocated within the 180 kHz bandwidth. In each new sub-channel selection, the transmitter generates a random number in a specific range named the Reselection Counter (RC). This number decreases by one in each broadcast, and resource reselection is needed when the RC value reaches zero. Since vehicle communications are concerned with passenger safety, minimizing delay in resource selection and transmission latency is critical to avoid cases such as those investigated in [18]. Therefore, 3GPP proposed the maximum allowed latency not to exceed 100 ms (in this work, it is denoted by T step ) [Table A. 1.5-1 of [20]]. High channel load can prevent vehicles from transmitting CAM messages due to the maximum allowed latency. RC values are generated by each vehicle in three ranges [5 − 15], [10 − 30], and [25 − 75], when T step is equal to 100, 50 and 20 ms, respectively [7]. In this work, 100 ms T step with [5 − 15] RC range is proposed as the reselection parameter.
The number of resource blocks in each subchannel depends on the amount of data to be sent and the selected modulation and coding scheme (MCS). The number of sub-channels in one subframe of 1 ms is given by the following expression: where RB Tot and RB SCH are the total number of resource blocks in a given bandwidth and the number of resource blocks in a signal subchannel, respectively, while the symbol ⌊x⌋ denotes the highest integer smaller than or equal to x. The total number of subchannels in T step is denoted as RES max and can be found as follows: The Channel Busy Ratio (CBR) is assumed in this work as a ratio of the used subchannels to the total number of subchannels in interval T step as follows [22]: where S busy and S sf are the number of radio resources that experience a signal-to-interference plus noise ratio γ higher than the minimum allowed signal-to-interference plus noise ratio γ min , and the number of all subchannels in the same subframe, respectively. Furthermore, assume that the received power and the transmitted power are denoted P r and P t , respectively. Let the path loss exponent be denoted as α, and the path loss coefficient (on the linear scale) at the reference distance 1 m be PL loss . The distance between the transmitter and the receiver is denoted by D. Then the received power can be written as follows: where G r is the gain of the receiving antenna. The distance D should be shorter than the maximum transmission range D max . Moreover, D max is related to transmitted power, and γ min by the following expression [22], [23]: where N 0 is the noise power in a subchannel. γ min can be obtained by the following steps: Step 1: Calculate the coding rate C rate for the desired amount of data N Bits [bits] as follows [23]: (6) where N symb , N SC , RB SCH indicate the number of symbols that carry the desired data (equal to 9), the number of subcarriers in RB (equal to 12), and the number of resource blocks, respectively. The number of bits per symbol is indicated by N el .
Step 2: Calculate the spectrum efficiency where N Tot , T Subf , and RB band are the total number of symbols in a subframe (equal to 14), the time duration of a sub-frame (equal to 1 ms), and the bandwidth of RB (equals 180 kHz), respectively.
Step 3: Calculate the minimum signal-to-interference-plusnoise ratio γ min as where is the expected implementation loss, which is assumed to be 0.6 [24].
In this work, two modulation and coding schemes are applied; thus, two values of γ min are calculated to obtain different D max according to the transmission power for each of them.

III. SYSTEM MODEL A. PROPOSED ALGORITHM
The proposed algorithm aims to generate multilevel CBR according to the concept of the DCC mechanism. The reaction of the system to the adjustment of CBR is represented by three factors at each level; maximum and minimum CBR, transmission power, and the established S-RSSI threshold. Moreover, the proposed channel load adjustment algorithm supports more flexible subchannel selection (with equal or lower than the maximum allowed schedule latency) by decreasing the number of vehicles in a vehicle transmission coverage range. This number of vehicles in a vehicle coverage range decreases by increasing the minimum threshold of SINR, which is equivalent to decreasing the receiving sensitivity and reducing the transmitted power based on CBR. In other words, the proposed algorithm increases or decreases channel utilization by tuning the minimum SINR threshold to the active levels of the DCC. The relaxed state is characterized by a minimum threshold for signal reception sensitivity that is represented by S-RSSI thr,min , maximum transmission power P t,max , and maximum acceptable CBR (CBR max,0 ). CBR max,0 is equal to the difference between unity and a specific fraction of CBR △CBR (0 < △CBR < 1) set by the operator. The active state consists of N active levels (n = 1, . . . , N ). Each level n has a maximum and minimum CBR, transmission power P t,n , and S-RSSI thr,n . In the restrictive state, S-RSSI , it is proposed that the lowest limit of CBR reaches 0.1 of CBR max,0 , while the transmission power reaches its lowest level.
The maximum CBR in the relaxed state can be expressed as follows.
where A is the maximum CBR in the relaxed state. The maximum and minimum CBRs in the active state n are assumed to be given by the formulae: CBR min,n = A n+1 CBR max,0 According to the above equations, the value of CBR for a given level is in the range between the highest and lowest CBR of this level. These boundaries have the following VOLUME 11, 2023 relationship with CBR max,0 of the relaxed state: The maximum and minimum CBR that can be applied in the restrictive state can be found as follows.
According to (10) and (12), the number of levels of CBR is inversely proportional to the value of △CBR (that is, when the value of △CBR increases, the number of levels decreases).
As an example of changing channel utilization in DCC access stages, see Fig. 1. The relaxed and restricted stages are the first and last stages in the sequence of DCC access state machine, whereas the stages between them reflect the active levels for which channel utilization varies. However, channel utilization reduction leads to a user performance increase and spectral efficiency improvement.
The procedure of the proposed algorithm is presented in the form of a pseudocode in Algorithm 1.
In the algorithm input, n represents the current TPC-DCC level that a vehicle has used in the last T step of the resource life cycle (when the RC reaches zero) before obtaining a new resource. CBR cur , and S-RSSI thr,n are the current CBR and the threshold of S-RSSI at the level n. A vehicle obtains CBR cur using (3), the subchannels are indicated as busy if S-RSSI is higher than S-RSSI thr,n of the current level. S-RSSI thr,n will continuously increase its threshold by 3 dB until it reaches the appropriate level of CBR. Furthermore, the number of increases also indicates the number of increases in levels. If the achieved number is greater than the number of levels, the next resource life cycle level will be the restrictive state. Furthermore, when CBR cur is lower than the minimum CBR of this level, S-RSSI thr,n will continuously decrease its threshold by 3 dB until it achieves a CBR compatible with the current maximum and minimum CBR. Suppose the zero level is achieved, then the state machine achieves the relaxed state. The number of decreasing steps will also be the decreasing number of levels. According to the new level, a new transmission power P t,n , the maximum and minimum CBR of the new level and a new S-RSSI thr,n will be established.

Algorithm 1 Proposed Adjusting Algorithm
Input: n, N , CBR cur , S-RSSI thr,n , CBR max,0 , A, RC Output: CBR max,n , CBR min,n , P t , S-RSSI thr,n , RC 1: while RC = 0 do 2: if CBR cur > A n CBR max,0 then 3: for i = 1 to N-n do 4: n = n + 1 5: S-SRRI thr,n = S-SRRI thr,n + 3 dB 6: CBR ← Calculate a new CBR using (3) 7: if CBR < A n CBR max,0 then 8: break for loop 9: end if 10: end for 11: else if CBR cur < A n+1 CBR max,0 then 12: for i = 1 to n do 13: if i ̸ = n then 14: S-SRRI thr,n = S-SRRI thr,n − 3 dB 15: CBR ← Calculate a new CBR using ( 25: CBR max,n ← A n CBR max,0 26: CBR min,n ← A n+1 CBR max,0 27: P t ← P t,n (P t assigned to level n (see Fig.2 The status of the sensed resources in the physical layer should be constantly reported to the data link layer for analysis. When RC reaches zero in subframe t, the MAC layer will ask the lower layer to send a report. This report should contain information on free resources in the period between [t − 1, (t − 10T step )]. The steps of S-SPS can be briefly summarized according to [7] as follows: • The resources in the subframe that would serve the vehicle V x are known as contiguous subchannels within the time interval [t + T 1, t + T 2] where T 1 can take any value lower than or equal to the time period of four subframes, depending on the processing time required for the selection operation. T 2 is the maximum CAM interval [20 ≤ T 2 ≤ 100], depending on the maximum allowed latency.
• The vehicle V x monitors all subframes in the time interval [t − 1, (t − 10T step )].
• Vehicle V x creates two lists: ListA that contains all resources that were sensed in the last time interval in the previous step, and ListB that is created as an empty list.
• In turn, V x excludes all candidates that have Reference Signal Received Power (RSRP) greater than the established threshold.
• The remaining candidates should be greater than or equal to 20% of all candidates before excluding them. Otherwise, the previous step will be repeated with the established threshold of RSRP increased by 3 dB.
• ListB is reported to the higher layer. The MAC layer randomly picks up the next resource from ListB.

C. EXTENDED-ESTIMATION AND RESERVATION OF RESOURCE ALLOCATION (E-ERRA)
E-ERRA is an algorithm proposed in [10] and [11] by the authors of this article as an alternative scheduling algorithm to S-SPS to improve the reliability of the selection of free resources and to avoid packet collision in C-V2X Mode 4. As we have already mentioned, this algorithm uses extra information included in the SCI (about 2 bytes). It contains the RC of a broadcast packet, one bit that is set when the resource location is reserved by another vehicle in the network, the location of the collided resources in the awareness range, and the next resource location to use when the RC reaches zero. Depending on the decoded SCI of the received packet, the E-ERRA algorithm performs the following steps.
• The physical layer of V x is continuously monitoring the subchannels of the last CAM interval [t − 1, t −T step ].
• ListA is initialized to collect all busy resources in the last CAM duration cycle T step .
• ListB is initialized to collect all resources in the last CAM duration cycle.
• The vehicle V x excludes all subchannel candidates from ListA that meet the following conditions:

1) They have an RC value that is not equal to the RC of V x . 2) They have Reference Signal Received Power
(RSRP) higher than the established threshold.

3) They have been indicated as resources reserved by
other vehicles in the awareness range (b resv ̸ = 0), where b resv is the bit indicating a location reserved by another vehicle. 4) They have been indicated as a collided resource.
• ListB collects all resources that meet the following conditions: 1) They are contained in subframes that have Reference Signal Received Power (RSRP) lower than the established threshold. 2) They are contained in subchannels that have not been indicated as reserved resource locations by other vehicles (b resv ̸ = 0).
• The remaining candidates for the resource locations in ListA should be placed in the First-In-First-Out (FIFO) queue with respect to receiving time. Furthermore, the remaining candidates in ListB should be arranged in the form of a resource location queue in ascending order with respect to the minimum S-RSSI.
• Vehicle V x points the first resource location in ListA as the next reselection resource location in its SCI, when the RC reaches five. This way, vehicles in the surrounding area will know the location of the reserved resource. The vehicle that uses the reserved location will activate the reservation bit (b resv = 1) at the next broadcast moment.
• Vehicle V x continuously checks the availability of the reserved subchannel resource in the awareness range to ensure that it will be available when reselection is carried out. Otherwise, the first location of the resource in the ListB queue will be indicated as the next resource location of V x .
• When the RC of V x reaches zero, the reserved resource location is achieved and a new random RC value is generated for the new allocation of resources. The RC generator should exclude the RC values of the packets transmitted through subchannels located in the same subframe to avoid overlapping in future resource reselection.

IV. SIMULATION SCENARIOS AND ASSUMPTIONS A. SYSTEM ASSUMPTIONS
A system layer simulation has been developed to test and compare the performance of the E-ERRA and S-SPS scheduling algorithms in interaction with the proposed multilevel adjustment of CBR in C-V2X Mode 4. The evaluation of the system model can be summarized in the following steps: 1) Each vehicle in the network evaluates the channel load at the last T step of its resource life cycle before resource reselection. 2) Two separate simulations are run with two different MCSs for a constant inter-vehicle distance (the density VOLUME 11, 2023 is constant over time) and for time-varying vehicle density. 3) The system is modeled to calculate the packet reception ratio and the resource collision ratio with both scheduling methods. 4) In (5), the maximum transmission distances are calculated for different transmission powers and the minimum signal-to-interference-plus-noise ratio is calculated for each MCS using (8). It is important to note that the number of vehicles in the awareness range is calculated in each CAM period for each vehicle in the network.

B. SIMULATION SCENARIOS AND SETTINGS
Different highway scenarios have been designed using the Matlab platform. Matlab was integrated with the open source Simulation of Urban MObility (SUMO) package [34] to model more realistic environments. The scenarios of this work have been classified into two groups: standard network and hybrid network. Each of these groups consists of two scenarios. These scenarios differ in terms of vehicle distribution in the network, transmission power, and signal reception sensitivity as follows: 1) Standard network: The performance of the system layer has been investigated in a fixed inter-vehicle distance and service range. Furthermore, the number of vehicles in each lane of the highway is constant (wrapped network) during simulation time, as proposed in [33]. As mentioned above, the awareness range depends on the transmission power. Scenarios 1 and 2 have been simulated for different transmission powers and different MCSs; see Table 2. 2) Hybrid network: In these scenarios, the performance of the system layer has been investigated with varying vehicle densities (the distances between vehicles change over time). Thus, the transmission power will correspond to CBR of the resources in the service range of each vehicle. The service range can vary from one vehicle to another on the same network. The number of vehicles moving in each lane varies over simulation time. Furthermore, the transmission power should be related to the channel load in the service range according to the proposed multilevel adjustment of CBR. Scenarios 3 and 4 have been evaluated for different transmission powers and different MCS schemes, and are parameterized in Table 2. The purpose of these scenarios is to test the performance of the proposed congestion control algorithm in different vehicle densities in the same network and the interaction between the resource allocation methods and the proposed algorithm in more realistic environments. In each simulation, the positions and distances of all vehicles in the network were updated and calculated in each CAM interval (assuming to be equal to 100 ms). Additionally, scheduling algorithms have been investigated and evaluated in all vehicles in the network for every CAM period. The common parameters for all scenarios are presented in Table 1. We assumed that the given channel bandwidth is 10 MHz. The Winner II B1 [33] propagation model was applied in our simulations. The maximum distances of the service ranges for specific transmission powers were calculated according to (5). Four highway scenarios were applied in several simulations that differed in selected parameters to obtain different results in the packet reception ratio and the packet collision ratio for different vehicle densities.

A. SYSTEM PERFORMANCE
In our simulations, we estimated two basic measures of packet transmission quality, namely, the Packet Reception Ratio PRR and the Collision Ratio CR. PRR is the ratio between the number of successfully decoded packets and the number of transmitted packets. CR is the ratio of packets that collided with other packets in the same coverage area to the total number of transmitted packets. PRR and CR have been calculated for all vehicles in the simulated network for each period of the CAM cycle for both resource scheduling algorithms in each scenario.
In order to be convinced that the simulation results obtained are sufficiently precise, we calculated the 95% confidence intervals for each estimated value applied in the figures showing PRR estimates. The width of each confidence interval [x − ε, x + ε] depends on the estimated value x, but the relative value of the tolerance ε/x does not basically exceed 7% or is often lower. However, to ensure a clear view of the estimated plots, we decided not to show the confidence intervals in the figures.

B. SIMULATION RESULTS
This subsection shows the performance of the system layer for two scheduling algorithms with the proposed algorithm to   adjust CBR levels. The E-ERRA and S-SPS scheduling algorithms have been tested, evaluated, and compared in two types   of TPC-DCC and adaptation of signal reception thresholds according to the calculation of the proposed levels. The performance of E-ERRA in terms of PRR in Fig. 3 for Scenario 1 (see Table 2) has shown promising results compared to S-SPS. PRR values increased from 11% over a distance of 100 m for all transmission powers to about 30-35% over 250 m and 2-20 % over 450 m for several transmission powers (see the legend in Fig. 3). In Scenario 2 (Fig. 5), the performance of PRR increases by about 18% for all transmission powers over a 100 m distance to 25-30% over 250 m, and to 11-30% in 450 m. The highest performance improvement for E-ERRA is observed when the highest transmission power is applied (in the relaxed state). Furthermore, high compatibility with the E-ERRA algorithm is observed when the proposed CBR levels are applied with the proposed adjustment method using TPC-DCC and adaptive sensitivity of reception signals.
Increasing CBR levels provides greater precision in resource selection. However, Fig. 4 of Scenario 1 and Fig. 6 of Scenario 2 have shown an exceptionally significant improvement in the packet collision ratio CR when the E-ERRA algorithm is applied. The E-ERRA algorithm that uses reservation and tracking resources extremely reduces packet collisions. The results have shown a 50-60% improvement for E-ERRA compared to S-SPS in all transmission powers. It should be stressed that the E-ERRA algorithm has very high performance stability, high reliability in selecting resources, and low complexity in processing and implementation.
The numbers of vehicles in Scenarios 3 and 4 vary over time during the simulation time. The proposed methods to adjust the parameters have motivated us to avoid resource collision by continuously calculating CBR and selecting the appropriate level of parameters of the physical layer. Transmission power changes appropriately according to the channel load of vehicles in the transmission area. Thus, in the same network, there are different service ranges. Fig. 7 of Scenario 3 and Fig. 8 of Scenario 4, presented the variability of the number of vehicles in the network. These vehicles  broadcast packets with different transmission powers but apply the same MCS in each scenario.
Figs. 9 and 10 showed the PRR performance for vehicles that use different transmission power for broadcasting. Fig. 9 shows that the system performance of the E-ERRA algorithm is higher by about 13-18% than the performance of S-SPS.
The effect of limited resources and the impact of interference on these vehicles in the same network when using QPSK modulation (in Scenario 3) appeared strongly in the packet reception ratio compared to the standard network. Furthermore, the efficiency of the proposed method of adjusting physical parameters is limited by limited resources.
On the other hand, in Fig. 10, in 16QAM applications (Scenario 4), the resources have a higher level of channel load than in Scenario 3, and more flexibility is possible in the selection between these resources (the high number of resources provided more flexibility in the selection between several levels of the proposed CBR levels). Furthermore, these features were strongly supported by using   the ListB queue in the E-ERRA algorithm (Section II-C) to avoid resource selection collisions and compensate for losses in reserved resources that can appear when owners of the reserved resources have left the awareness range. In this scenario, E-ERRA has shown gradual enhancement compared to S-SPS by up to 27% when the distance is 300 m. The packet collision ratio in Figs. 11 and 12 in E-ERRA has shown about 40% higher efficiency compared to S-SPS by avoiding packet collision. Consequently, E-ERRA is more convenient than S-SPS broadcasting in the hybrid network environment. Increased levels of CBR with TPC-DCC and adaptive signal reception sensitivity show high compatibility with the E-ERRA algorithm compared to S-SPS.

VI. CONCLUSION
In this article, the DCC challenges for autonomous resource selection in V2X are investigated when channel load is high in the network. In this article, TPC-DCC and an adaptive power threshold of the received signal are engaged with a proposed algorithm of channel load adjustment based on the channel busy ratio in order to reduce the collisions in packets/resources and the interference effect. The E-ERRA and S-SPS resource allocation algorithms are considered for this purpose. The interactions of the proposed solutions with S-SPS and E-ERRA are investigated for fixed and flexible inter-vehicle distance environments. The simulation results for standard networks with E-ERRA supported by TPC-DCC have shown higher PRR, reliability in radio resource selection, and lower CR than S-SPS for all levels of physical parameters.
The E-ERRA algorithm is applied with TPC-DCC, adaptive signal sensitivity during reception, the proposed multilevel channel busy ratio algorithm in a hybrid network, and a variable number of vehicles in the same broadcast range. These network features have improved performance stability, PRR, and immunity against packet losses and CR compared to the case where the S-SPS algorithm is applied.