GEORGIA DOT RESEARCH PROJECT 20-03 Final Report Adequacy of DSRC in 5.9 GHz Band for GDOT's Connected Vehicle Infrastructure Office of Performance-based Management and Research 600 West Peachtree Street NW Atlanta, GA 30308 TECHNICAL REPORT DOCUMENTATION PAGE 1. Report No.: 2. Government Accession No.: FHWA 21-2003 N/A 4. Title and Subtitle: Adequacy of DSRC in 5.9 GHz Band for GDOT's Connected Vehicle Infrastructure 7. Author(s): Seungmo Kim, Ph.D. 3. Recipient's Catalog No.: N/A 5. Report Date: May 2021 6. Performing Organization Code: 6250002291 8. Performing Organization Report No.: 20-03 9. Performing Organization Name and Address: Georgia Southern University Research and Service Foundation, Inc 261 Forest Drive Statesboro, GA.30458-8005 10. Work Unit No.: N/A 11. Contract or Grant No.: PI# 0016970 12. Sponsoring Agency Name and Address: 13. Type of Report and Period Covered: Georgia Department of Transportation Final; May 2020 May 2021 Office of Performance-based Management and Research 600 West Peachtree St. NW 14. Sponsoring Agency Code: N/A Atlanta, GA 30308 15. Supplementary Notes: Conducted in cooperation with the U.S. Department of Transportation, Federal Highway Administration. 16. Abstract: Vehicle-to-everything (V2X) communications are expected to take a critical role in a variety of transportation safety applications in connected and autonomous vehicle environment. However, Dedicated Short-Range Communications (DSRC), one of the representative technologies implementing the V2X communications, is encountering challenges due to the re-allocation of the 5.9 GHz spectrum by the Federal Communications Commission (FCC). The state of Georgia is leading the nation in the deployment of connected vehicle infrastructure with thousands of infrastructure units connecting vehicles based on DSRC. This project aimed at measuring the impact of the spectrum re-allocation on the performance of the Georgia's connected vehicle infrastructure. The project's particular technical focuses were to (i) model the performance of a DSRC system and (ii) design a protocol to improve the DSRC performance, under the FCC's reform of the 5.9 GHz band. 17. Keywords: Connected vehicles, 5.9 GHz band reform, DSRC, U.S. FCC 18. Distribution Statement: No Restriction 19. Security Classification 20. Security Classification (of this (of this report): page): Unclassified Unclassified 21. No. of Pages: 22. Price: 56 Free Form DOT 1700.7 (8-72) Reproduction of completed page authorized. GDOT Research Project 20-03 Final Report ADEQUACY OF DSRC IN 5.9 GHZ BAND FOR GDOT'S CONNECTED VEHICLE INFRASTRUCTURE By Seungmo Kim, Ph.D Assistant Professor, Department of Electrical and Computer Engineering Georgia Southern University Contract with Georgia Department of Transportation In cooperation with U.S. Department of Transportation Federal Highway Administration May 2021 The contents of this report reflect the views of the authors who are responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official views or policies of the Georgia Department of Transportation or the Federal Highway Administration. This report does not constitute a standard, specification, or regulation ii Symbol in ft yd mi in2 ft2 yd2 ac mi2 fl oz gal ft3 yd3 oz lb T oF fc fl lbf lbf/in2 SI* (MODERN METRIC) CONVERSION FACTORS APPROXIMATE CONVERSIONS TO SI UNITS When You Know Multiply By To Find LENGTH inches 25.4 millimeters feet 0.305 meters yards 0.914 meters miles 1.61 kilometers AREA square inches 645.2 square millimeters square feet 0.093 square meters square yard 0.836 square meters acres 0.405 hectares square miles 2.59 square kilometers VOLUME fluid ounces 29.57 milliliters gallons 3.785 liters cubic feet 0.028 cubic meters cubic yards 0.765 cubic meters NOTE: volumes greater than 1000 L shall be shown in m3 MASS ounces 28.35 grams pounds 0.454 kilograms short tons (2000 lb) 0.907 megagrams (or "metric ton") TEMPERATURE (exact degrees) Fahrenheit 5 (F-32)/9 Celsius or (F-32)/1.8 ILLUMINATION foot-candles foot-Lamberts 10.76 3.426 lux candela/m2 FORCE and PRESSURE or STRESS poundforce 4.45 newtons poundforce per square inch 6.89 kilopascals Symbol mm m m km mm2 m2 m2 ha km2 mL L m3 m3 g kg Mg (or "t") oC lx cd/m2 N kPa Symbol mm m m km mm2 m2 m2 ha km2 mL L m3 m3 g kg Mg (or "t") oC lx cd/m2 N kPa APPROXIMATE CONVERSIONS FROM SI UNITS When You Know Multiply By To Find millimeters meters meters kilometers LENGTH 0.039 3.28 1.09 0.621 inches feet yards miles AREA square millimeters 0.0016 square inches square meters 10.764 square feet square meters 1.195 square yards hectares 2.47 acres square kilometers 0.386 square miles VOLUME milliliters 0.034 fluid ounces liters 0.264 gallons cubic meters 35.314 cubic feet cubic meters 1.307 cubic yards MASS grams 0.035 ounces kilograms 2.202 pounds megagrams (or "metric ton") 1.103 short tons (2000 lb) TEMPERATURE (exact degrees) Celsius 1.8C+32 Fahrenheit ILLUMINATION lux candela/m2 0.0929 0.2919 foot-candles foot-Lamberts FORCE and PRESSURE or STRESS newtons 0.225 poundforce kilopascals 0.145 poundforce per square inch Symbol in ft yd mi in2 ft2 yd2 ac mi2 fl oz gal ft3 yd3 oz lb T oF fc fl lbf lbf/in2 *SI is the symbol for the International System of Units. Appropriate rounding should be made to comply with Section 4 of ASTM E380. (Revised March 2003) iii Table of Contents List of Figures..................................................................................................................... vi List of Tables .....................................................................................................................vii List of Abbreviations and Symbols ..................................................................................viii Executive Summary............................................................................................................. 1 1. Introduction ..................................................................................................................... 2 1.1. Background ............................................................................................................... 2 1.2. The FCC's 5.9 GHz Band Reform ............................................................................ 3 1.3. Possible Bandwidth Contention with C-V2X ........................................................... 4 1.4. Literature Review...................................................................................................... 4 1.4.1. Stochastic Methods for Modeling the Performance of DSRC ........................... 4 1.4.2. DSRC Performance Measurement Metrics ........................................................ 6 1.4.3. Inter-Technology Coexistence in the 5.9 GHz Band.......................................... 7 2. Preliminary ...................................................................................................................... 8 2.1. Background of DSRC ............................................................................................... 8 2.2. Multiple Access in DSRC ......................................................................................... 9 2.3. Outcomes from Attempt of BSM Transmission ..................................................... 10 3. Analysis ......................................................................................................................... 13 3.1. Mathematical Modeling of CSMA/CA in DSRC ................................................... 13 3.2. Calculation of DSRC Performance ......................................................................... 15 3.2.1. Ptx: Probability of BSM Transmission at Vehicle ............................................ 15 3.2.2. Pexp: Probability of Expiration .......................................................................... 18 3.2.3. Pcol: Probability of BSM Collision Over-the-Air.............................................. 19 3.2.4. PDR: Packet Delivery Rate .............................................................................. 22 4. Experiments ................................................................................................................... 24 5. Results ........................................................................................................................... 27 5.1. General 2-D Setting................................................................................................. 27 5.1.1. Ptx ...................................................................................................................... 29 5.1.2. Pcol ..................................................................................................................... 31 5.1.3. PDR................................................................................................................... 33 5.2. Urban Canyon Scenario .......................................................................................... 36 iv 5.2.1. Key Performance Determining Factor.............................................................. 36 5.2.2. DSRC Performance........................................................................................... 37 5.3. Suburban Highway Scenario................................................................................... 41 5.3.1. Key Performance Determining Factor.............................................................. 41 5.3.2. DSRC Performance........................................................................................... 41 6. Conclusions ................................................................................................................... 43 References ......................................................................................................................... 45 v List of Figures Figure 1-1. Illustration. FCC's proposed 5.9 GHz band reform ..................................................... 3 Figure 2-1. Illustration. An example scenario of HN ................................................................... 10 Figure 4-1. Illustration. Structure of the testbed implementing a DSRC Tx ................................ 23 Figure 4-2. Illustration. Flowchart and structure of the simulation software ............................... 26 Figure 4-3. Screenshot. Distribution of vehicles on a general 2-D geographical setting ............. 27 Figure 4-4. Graph. Ptx versus Nsta according to CW in a general 2-D setting............................... 29 Figure 4-5. Graph. Pcol versus Nsta according to CW in a general 2-D setting ............................. 31 Figure 4-6. Graph. PDR versus Nsta according to CW in a general 2-D setting ........................... 33 Figure 4-7. Screenshot. A snapshot of the urban canyon scenario (100 vehicles (small rectangles 2 m-by-7 m in size) and 200 buildings (grey squares, 50 m-by-50 m in size) uniformly randomly distributed on a 6-lane, 4-way junction in a 1 km-by-1 km two-dimensional space)................... 35 Figure 4-8. Screenshot. A snapshot of the suburban highway scenario (100 vehicles uniformly randomly distributed on a 4-lane, 4-way junction in a 1 km-by-1 km two-dimensional space) .. 39 Figure 4-9. Graph. Comparison of DSRC performance (in terms of PDR) between urban and suburban scenarios (with CW = {31, 127}).................................................................................. 40 vi List of Tables Table 4- 1. Blockage rate according to vehicle density ................................................................ 37 vii List of Abbreviations and Symbols Abbreviation / Symbol BSM CCH CSMA/CA C-V2X CV CW DCF DIFS DLVY DSRC EIFS EXP FPGA GDOT HN ITS ncs ntx Pb Pini Pcol PDR Psync Ptx QPSK RAT Rx SCH SDR STA SYNC Meaning Basic safety message Control channel Carries-sense multiple access with collision avoidance Cellular vehicle-to-everything communications Connected vehicle Contention window Distributed coordination function DCF interframe space Successful BSM delivery to a receiver vehicle Dedicated short-range communications Extended interframe space Expiration of a BSM as result of failing to transmit within 100 msec Field-programmable gate array Georgia Department of Transportation BSM collision due to hidden nodes Intelligent transportation system Number of other vehicles detected within a vehicle's carrier-sense range Number of other vehicles actually transmitting a BSM Probability that a vehicle finds the channel busy Probability of a vehicle choosing an initial backoff value Probability that a sent BSM experiences an over-theair collision with a BSM sent by another vehicle Packet delivery rate: Probability that a BSM is successfully delivered to a receiver vehicle Probability of a synchronized transmission Probability that a vehicle is able to make it through a backoff process and sent a BSM Quadrature phase shift keying Radio access technology Receiver Shared channel Software-defined radio Station (e.g., vehicle, infrastructure, pedestrian, etc.) transmitting signals BSM collision due to a synchronized transmission viii Tx USRP V2X VOI Transmitter Universal Software Radio Peripheral Vehicle-to-everything communications Vehicle of interest ix Executive Summary Vehicle-to-everything (V2X) communications are expected to take a critical role in a variety of transportation safety applications in connected and autonomous vehicle environment. However, Dedicated Short-Range Communications (DSRC), one of the representative technologies implementing the V2X communications, is encountering challenges due to the re-allocation of the 5.9 GHz spectrum by the Federal Communications Commission (FCC). The state of Georgia is leading the nation in the deployment of connected vehicle infrastructure with thousands of infrastructure units connecting vehicles based on DSRC. This project aimed at measuring the impact of the spectrum re-allocation on the performance of the Georgia's connected vehicle infrastructure. The project's particular technical focuses were to (i) model the performance of a DSRC system and (ii) design a protocol to improve the DSRC performance, under the FCC's reform of the 5.9 GHz band. This project has produced the following outcomes: Theoretical analysis and network-level computer simulations were established. Performance of DSRC were evaluated, getting ready to apply the framework to comparison with other coexisting technologies, e.g., Wi-Fi. A general inter-technology coexistence scenario was developed, and its expected performance was evaluated. Particular focus was on evaluation of the reliability rather than the latency, considering that the likely use cases are related to the traffic safety. 1 1. Introduction 1.1. Background Vehicle-to-Everything (V2X) communications have the potential to significantly bring down the number of vehicle crashes, thereby reducing the number of associated fatalities (US Department of Transportation 2017). The capability gave V2X communications the central role in the operation of intelligent transportation systems (ITSs) for connected vehicle environments. Today, the two main radio access technologies (RATs) that enable V2X communications are Dedicated Short Range Communications (DSRC) and Cellular-V2X (C-V2X). The DSRC is designed to operate only in the 5.9 GHz band (5.850-5.925 GHz), which has been earmarked in many countries for ITS applications. On the other hand, C-V2X can operate not only in the 5.9 GHz band but in the cellular operators' licensed bands as well (Wang, Mao and Gong 2017). Between the two RATs, DSRC has longer been tested in many countries for safety-critical applications. As such, the most important benefit of DSRC is that it is a proven technology for a longer time by many different organizations. Furthermore, DSRC is not bounded by patents, which requires no telecom subscription to use it. To take these advantages, since the 5.9 GHz band was dedicated for DSRC in the United States (U.S.) by the Federal Communications Commission (FCC) in 1999, as of November 2018, more than 5,315 roadside units (RSUs) operating in DSRC were deployed nationwide (Kenney Nov. 2018). In December 2016, the National Highway Traffic Safety Administration (NHTSA) proposed to mandate DSRC for all new light vehicles (NHTSA 2016). However, despite the advantages and widespread 2 Figure 1-1. Illustration. FCC's proposed 5.9 GHz band reform deployment, there are several key issues to resolve in order to guarantee stable operations of DSRC. 1.2. The FCC's 5.9 GHz Band Reform Out of the 75 MHz of bandwidth in the 5.9 GHz band (i.e., 5.870-5.925 GHz), in December 2019, the U.S. FCC voted to allocate the lower 45 MHz (i.e., 5.850-5.895 GHz) for unlicensed operations to support high-throughput broadband applications (e.g., Wireless Fidelity, or Wi-Fi) (US FCC Dec. 2019.). Figure 1-1 illustrates this proposed bandwidth reform. The key points of the reform are two-fold: (i) leaves the upper 30 MHz (i.e., 5.895-5.925 GHz) for ITS operations; (ii) dedicates the upper 20 MHz of the chunk (i.e., 5.905-5.925 GHz) for C-V2X. According to this plan, DSRC is only allowed to use 10 MHz of spectrum (i.e., 5.895-5.905 GHz). It has never been studied nor tested if 10 MHz would suffice for the operation of the existing DSRC-based transportation safety infrastructure. The state of Georgia is a recognized national leader in the deployment of connected vehicle infrastructure, which also operates in DSRC. As such, it has become urgent to understand how much impact the FCC's 5.9 GHz band reform will be placed on the performance of Georgia's connected vehicle infrastructure. 3 1.3. Possible Bandwidth Contention with C-V2X According to the FCC's 5.9 GHz band reform (US FCC Dec. 2019.), DSRC may need to coexist with cellular vehicle-to-everything (C-V2X) users in the upper 30 MHz (i.e., 5.895-5.925 GHz). The key technical challenge here is that C-V2X uses a different technology standard, and hence the technology is incompatible with DSRC-based operations. Based on the principal investigator (PI)'s recent investigation (Kim and Bennis 2019), on average 4.63 C-V2X vehicles can corrupt a 10-MHz channel of DSRC. It implies that coexistence with 5 C-V2X vehicles may significantly degrade the performance of a DSRC system. 1.4. Literature Review 1.4.1. Stochastic Methods for Modeling the Performance of DSRC Carrier sense multiple access with collision avoidance (CSMA/CA) for the general Institute of Electrical and Electronics Engineers (IEEE) 802.11 has been modeled as a two-dimensional Markov chain (Bianchi 1998). However, in DSRC, the contention window size does not get doubled even upon a packet collision, which simplifies the Markov chain structure. The key difference in our model is that for basic safety message (BSM) broadcast in a DSRC system, the probability of decrementing a backoff counter is not 1. Also, a Markov chain has been proposed to model on the broadcast of safety messages in DSRC (Yin, Ma and Trivedi 2013); yet it does not take into account a packet expiration (EXP) during a backoff process. Further, there was another aspect about which the model did not describe completely accurately. According to the IEEE 802.11p-2010 standard (IEEE 2010), the fundamental access method of 4 the IEEE 802.11 medium access control (MAC) is a distributed coordination function (DCF), which shall be implemented in all stations (STAs). Investigating the CSMA/CA written in IEEE 802.11 MAC (Bianchi 1998) (IEEE 2016), the probability of transmission at the state of backoff being 0 still requires the probability of 1 - Pb for a transmission (where Pb denotes the probability of a slot being busy). Also, the previous models suggested for analyzing the 802.11p beaconing have paid little attention to the varying number of contending nodes (Yao, Rao and Liu 2013) and the restricted channel access of the control channel (CCH), i.e., Channel 178 (5.885-5.895 GHz) (Khabazian, Aissa and Mehmet-Ali 2013). Since the 802.11p MAC protocol is a contention-based scheme, the joint effect of the varying number of contending nodes and the restricted channel access may lead the network to perform quite differently (Lei and Rhee 2019). However, Ptx, the probability of a vehicle's making it through a backoff process, is given too simple in (Lei and Rhee 2019), which limits the model's applicability. That is, while DSRC usually adopts an exponential backoff, the model overly simplified the probability as Ptx = 2/(CW+1), which can only be used when there is no exponential backoff (Bianchi 2000). Notice that CW stands for a "contention window," from which a STA chooses a backoff counter value. Given the significance of an EXP in determining the performance of a DSRC system, the models cannot be considered to completely accurately characterize the behavior of a safety message broadcast. In this work, we develop a new mathematic model that integrates those two factors. 5 1.4.2. DSRC Performance Measurement Metrics Packet delivery rate (PDR) is one of the classical metrics that are used to measure the performance of a communications network. The PDR is defined as the ratio of data packets that are actually received at the receiver end to those which were originally sent by the sender. Besides the classical metrics, some other metrics have also been used to evaluate the performance of congestion control techniques. Examples include the probability of successful reception of beacon message (Torrent-Moreno, et al. 2009), update delay (Kloiber, et al. 2015) as the elapsed time between two consecutive BSMs successfully received from the same transmitter, a 95% Euclidean cutoff error (in meters) (Huang, et al. 2010), and information dissemination rate (IDR) (Fallah, et al. 2010). Other metrics could be found in the literature as well. The inter-reception time (IRT) is a metric motivated from the need for displaying both successful and failed transmissions at once. A latest work proposed an algorithm that adapts the frequency of BSM broadcast according to the IRT (Son and Park 2019). The key limitation of the metric is that it measures the BSM reception performance at a single vehicle, which can provide a myopic view only. For that reason, this project adopts the PDR as the main metric in the evaluation of the performance of a DSRC network. In fact, compared to receiver vehicle-centric metrics, the PDR better serves our purpose of evaluating a system as an entirety rather than measuring the performance of a single vehicle. The key innovation of this project is that we provide an extensive mathematical analysis framework as well as a comprehensive computer simulation framework. This distinguishes our work from other simulation-based work (ElBatt, et al. 2006) 6 (Jornod, et al. 2019) (Almeida, et al. 2018) (Cheng, et al. 2017) (Khan, Hoang and Harri 2017) and experiment-based work (Liu, et al. 2016) (Rendaa, et al. 2016). 1.4.3. Inter-Technology Coexistence in the 5.9 GHz Band The coexistence problem among dissimilar RATs in the 5.9 GHz band has been discussed in the literature: (i) between DSRC and Wi-Fi (Kim 2017) and (ii) between DSRC and Wi-Fi/C-V2X (Kim and Bennis 2019). Modification of CSMA was proposed for relieving bandwidth contention among vehicles within an IEEE 802.11p network (Kim and Dessalgn 2019). The inter-vehicle distance was selected as the factor representing the risk of a crash. A message prioritization scheme among different classes of vehicles was proposed for military vehicles over commercial ones (Kim and Dessalgn 2019). The key limitation was a relatively simple model for the stochastic geometry: a single junction of two 6-lane road segments. Such an urban model may lose generality when applied to other scenarios. In another latest work, a reinforcement learning-based approach was proposed to address the dynamicity of a V2X networking environment (Kim and Kim 2020). Each vehicle needs to recognize the frequent changes of the surroundings and apply them to its networking behavior, which was formulated as a multi-armed bandit (MAB) problem. The MAB-based reinforcement learning enabled a vehicle, without any assistance from external infrastructure, to (i) learn the environment, (ii) quantify the accident risk, and (iii) adapt its backoff counter according to the risk. 7 2. Preliminary 2.1. Background of DSRC The Basic Safety Message (BSM) is one of a set of messages defined in the SAE Standard J2735, Dedicated Short Range Communications (DSRC) Message Set Dictionary. The BSM consists of two parts: Part 1 is sent in every BSM message and Part 2 consists of a large set of optional elements. Not all elements are available from all vehicles, and which elements are sent, if available, will be based on event criteria that are not specified in the J2735 standard. The BSM garners particular interest in the sense that a vast majority of safety-critical applications are based on the BSM. Therefore, this project focused on exchange of BSMs among vehicles. Moreover, mainly due to the U. S. FCC's latest decision to reduce the bandwidth for DSRC, this project investigated the performance of DSRC in the worst-case scenario. That is, although the 5.9 GHz band defines 7 channels, this project assumed that vehicles in a network is allowed to use a single channel only. It means that the result has a room for improvement if the network's channel selection is expanded among the other 6 channels. As such, the results that will be demonstrated in this report can be regarded as the most conservative ones. 2.2. Multiple Access in DSRC In telecommunications, a multiple access method is defined to allow multiple STAs connected to a same transmission medium (i.e., channel) over which they transmit their signals. That way, the selection of a multiple access method usually is a critical factor that defines the capacity of a telecommunications system. As an unlicensed technology that requires no monetary payment from the subscribers, the IEEE 802.11 standard adopts a multiple access method that each STA can implement without any assistance from a central entity. As such, the standard adopts carrier-sense multiple access (CSMA) as its primary scheme to manage multiple STAs attempting to access to a certain finite number of channels. More specifically, the standard usually implements the CSMA with collision avoidance (CSMA/CA) in which carrier sensing is used, but nodes attempt to avoid collisions by beginning transmission only after the channel is sensed to be "idle" (IEEE 2016). When they do transmit, nodes transmit an entire packet data. As a family of the IEEE 802.11 standard, DSRC is no exception in adopting CSMA/CA as its primary multiple access scheme. Through this project, this research team was able to model the CSMA/CA procedure based on Markov process for better visualization and intuition. Technical details on CSMA/CA for DSRC will be provided in Section 3.1. 9 Figure 2-1. Illustration. An example scenario of HN 2.3. Outcomes from Attempt of BSM Transmission As a result of an attempt of a BSM transmission at a vehicle, there can be four possible outcomes (Stanica, Chaput and Beylot 2014). Expiration (EXP): Although nominal yet, a vehicle transmits a packet (i.e., BSM) every 100 msec. As such, if a vehicle is unsuccessful to have a BSM transmitted within a window of 100 msec, the vehicle needs to discard the current BSM and proceed to the next one. This event of the vehicle's inability to transmit a BSM is defined as a BSM expiration. Collision by synchronized transmission (SYNC): Even if a BSM has been transmitted, the BSM can still go through a "collision" with (an)other BSM(s) over the air. There are two types of collision. The first type of collision is a collision caused by a "synchronized transmission." Recall that the way that DSRC deals with multiple access is to let each vehicle to randomly choose a backoff value within the range of [0, CW-1]. Since the random value selection process is completely standalone at each vehicle and not 10 coordinated by any central entity, it is possible that multiple vehicles select a same backoff value. Those vehicles having selected the same backoff value must start their transmissions at the same time, which inevitably yields a collision among their BSMs. This type of collision is called a synchronized transmission. Collision by hidden node(s) (HN): A BSM collision can also occur by (a) hidden node(s). A hidden node is defined as a BSM transmitting node that is outside of the vehicle-of-interest's carrier-sense range. That is, the vehicle of interest is not able to hear the hidden node's signaling. It likely yields that both the vehicle of interest and the hidden node transmit their BSMs, which leads to a collision of the BSMs. Figure 2-1 sketches this concept. Successful BSM delivery (DLVY): A successful BSM delivery is defined as two events occurring in a series: (i) transmission of a BSM without an EXP and (ii) no collision over the air, i.e., no SYNC nor HN. 11 Figure 3-1. Illustration. CSMA/CA for DSRC modeled as a Markov process 12 3. Analysis 3.1. Mathematical Modeling of CSMA/CA in DSRC Recall from Section 2.2 that DSRC adopts CSMA/CA as its multiple access method, inherited from the IEEE 802.11 standard. To summarize the procedure, if no medium activity is indicated for the duration of a particular backoff slot, then the backoff procedure shall decrement its backoff time by a slot time. If the medium is determined to be busy at any time during a backoff slot, then the backoff procedure is suspended; that is, the backoff timer shall not decrement for that slot. The medium shall be determined to be idle for the duration of a DIFS or EIFS, as appropriate, before the backoff procedure is allowed to resume. The first quantity to model a CSMA/CA process is Pini, the probability of a vehicle's selecting a backoff counter value. To elaborate, DSRC adopts the CSMA/CA in which the probability of being allocated a backoff counter is uniform in the range of [0, CW-1] where CW denotes the size of a contention window. We remind that a CW refers to a range of integers from which each vehicle picks up a value as its backoff counter for each BSM transmission. Before every BSM transmission, the vehicle decrements the backoff counter down until the counter reaches 0. Only then are stations allowed to transmit the frame if the medium is still idle. For instance, a vehicle having selected 0 as its backoff counter does not need to hold its transmission, while another vehicle having chosen 4 has to refrain from transmission until it has the counter decremented to 0. When a collision occurs, each vehicle doubles the contention window size to reduce the probability of a subsequent collision, up to a fixed maximum contention window size. This is called Binary Exponential Backoff (BEB). The initial small contention window size is referred to 13 as the minimum contention window (CWmin) and the capped maximum size is referred to as the maximum contention window (CWmax). The values for CWmin and CWmax are set depending on the type of data (Qiu, et al. 2014). Figure 3-1 illustrates the backoff process that a vehicle goes through before transmission of a BSM. Notice that Pini is a function of c denoting the value of the backoff counter that the vehicle has chosen, where the value ranges in {0,1, , CW - 1}. The first row of the diagram in Figure 3-1, Bc, indicates the current value of the backoff counter. (The name "B" indicates "backoff.") Hence, the first allocation of a Bc is solely determined by Pini. State B0 indicates the state where the vehicle has decremented the backoff counter to 0 within 100 msec. It implies that if the vehicle has chosen a large value of the backoff counter c, it has to make a longer process to reach B0. At each state Bc, in order to decrement the counter c, the vehicle has to observe the channel idle, which is indicated by 1 Pb where Pb denotes the probability that the channel is busy (i.e., not idle). When the vehicle finds the busy, it has to spend a time slot at the same state Bc before it observes the channel again in the next time slot. This delay at a backoff state Bc is expressed as a state Dc,d where d denotes the index for a delay and ranges in d {0,1, , }. Notice that denotes the maximum number of time slots that a vehicle can spend for delays at a backoff state Bc, which can be formally written as = - - (1) where Lbcn and lbcn denote the number of time slots defining the length of a beaconing period (i.e., 100 msec) and the length of a BSM, respectively. Note from the equation that -lbcn comes 14 from the length of a BSM itself, while -c is given from the number of backoffs that has to be decremented. It is noteworthy from the equation that gets larger with a smaller c, which is translated to that a vehicle assigned a smaller backoff counter c has a higher chance to reach B0. 3.2. Calculation of DSRC Performance Based on the analytical framework that was described in Section 3.1, we proceed to definition of metrics that will be used for evaluation of the performance of a DSRC network. 3.2.1. Ptx: Probability of BSM Transmission at Vehicle Let Ptx denote the probability that a vehicle transmits in any slot within a beaconing period (which is composed of Lbcn slots). In other words, Ptx gives the probability that a vehicle has been able to reach state B0 in the backoff process (that is shown in Figure 3-1) within a 100-msec beaconing period without experiencing an EXP. Definition of Ptx based on the Markov-process analysis takes the following calculations. We start from finding the transition matrix for the Markov process given in Figure 3-1. Let us denote the transition matrix by M, which is given by 15 (2) Based on the transition matrix M, we can proceed to calculating the probability of a one-step steady-state transition from two given states, i.e., Bc Bc-1. Let = lim [() = ] be the stationary distribution of the Markov chain to obtain the steady-state where B(t) denotes a backoff counter state at time t. A one-step transition between two arbitrary states (i.e., Bc Bc-1) in steady state for the Markov chain can be formulated as below: (3) 16 Notice from (a) that the probability of state Bc is composed of two parts: (i) directly from B0 and (ii) via Bc+1. As shown in (b), the same holds for Bc+1, from which one can infer the pattern of steady-state propagation for the Markov chain. Also, in (c), we substitute P[Bc Bc-1] with the probability of 1 -Pb as illustrated in Figure 3-1. Lastly, (d) tells that the expansion shown in the equation above goes until it reaches P[BCW-1]. It is important to note that state BCW-1 has only one input: directly from B0. Now, the probability of reaching state B0 can be formulated as a function of Pb. The probability is equivalent to the probability of a BSM transmission after propagating through the Markov chain (Bianchi 1998). The general one-step steady-state transition shown in the equation of -1 above can be applied to obtain the probability of transition from an arbitrary state Bc to B0. In this research, this transition formula was regarded to provide a piece for the numerical solve of Ptx and Pb, which is given by (4) Implication of the equation given above is that now Ptx is written as a function of Pb since all the Pini's are known. 17 Now, the other side of the numerical solve should be provided. While the Markov chain describes a node's process of going through a backoff process in CSMA/CA, Pb is determined from the dynamics among the nodes competing for the channel. More specifically, all the nodes that are located in the carrier-sense range of the tagged vehicle become the competitors for the channel. The probability of a slot being found busy can be formulated as = 1 - (1 - ) (5) where ncs denotes the number of nodes within a given node's range of carrier sensing. We compute Ptx and Pb via numerical solve based on simultaneous equations composed of the last two equations on Ptx and Pb, respectively. The simultaneous equations were to be solved via a sufficiently large number of iterations testing all the possible values for and Pb. 3.2.2. Pexp: Probability of Expiration Based on Ptx, one can proceed to obtaining the probability that a vehicle is not able to transmit in any slot within a beaconing period (i.e., 100 msec), which is given by Pexp = 1 - Ptx (6) To prove the formula, it is significant to note that a backoff process of DSRC is "reset" after every beaconing period regardless of whether a BSM is successfully transmitted or not. It means that the "steady-state" in DSRC is t Lbcn (Slot time) rather than as in other IEEE 18 802.11 standards. As such, cases of finishing at other states including B's and D's occur although a separate state of EXP exists in the backoff process model as shown in Figure 3-1. Recall from the figure that state ,, the deepest state of a delay with backoff counter c, gives the last possible delay state before being drained to an EXP. It implies the possibility of staying at any other state than B0 and EXP at the end of a beaconing period, which consequently should be added to the probability of an EXP. As a consequence, the probability of an EXP is obtained as Pexp = 1 - Ptx. 3.2.3. Pcol: Probability of BSM Collision Over-the-Air Once a BSM is transmitted, in order to lead to a successful delivery, no collision must occur on the BSM during the transit. In other words, for a vehicle of interest (VOI), an external collision occurs unless both of the following conditions are satisfied: (i) when the VOI is transmitting, no vehicles within its carrier-sensing range delivery BSMs at the same time slot; and (ii) when the VOI is transmitting, no hidden terminals should corrupt the VOI's BSM. To formulate, as just have mentioned, there are two types of collision in an IEEE 802.11-based networknamely, SYNC and HN (Stanica, Chaput and Beylot 2014). Notice that the geometry for a SYNC is limited to the VOI's carrier-sense range because the BSM collision occurs among the vehicles who are sensed but cannot avoid a collision due to allocation of a same backoff value (Kim and Bennis 2019). On the other hand, the geometry for a HN is greater than a SYNC because the vehicles causing this type of BSM collision are placed outside of the VOI's carriersense range (Kim and Bennis 2019), which figuratively defines "hidden nodes." 19 Based on the rationale, the probability that a BSM experiences a collision over the air can be formulated based on SYNC and HN as (6) Notice that Psync denotes the probability that a SYNC occurs on a BSM over the air, which can be formulated as Psync = [SYNC] = 1 - [No SYNC] (7) where [No SYNC] = ([ =0 (8) CW][ Tx vehicles][All different backoffs among them]) with denoting an Iverson bracket: [ CW] = {10,, CW otherwise (9) It indicates that in order for a SYNC not to occur, there must be only CW vehicles transmitting, which leaves a possibility that all vehicles are assigned all different backoff values. In other words, a SYNC can never be avoided when > CW vehicles compete for the channel. The next two terms in the equation of P[No SYNC] can be found as [ Tx vehicles] = () (1 - )- and 20 (10) [All different backoffs among them] All different backoff selections = All possible backoff selections by vehicles (11) = CW! [(CW - )!]-1 CW Now, we switch our attention to calculation Phn, which can be found based on a similar framework with that for Psync: Phn = [HN] = 1 - [No HN] (12) where [No HN] = ([ =0 (13) CW][ Tx vehicles][No hidden terminal among them]) with [ CW] and [ Tx vehicles] having already been computed as part of calculation of Psync. The only quantity that remains to be quantified is [No hidden terminal among them] = [(|C~W | ) ] (14) where ~ denotes the set of vehicles not causing a HN. We note that the set depends on which backoff value c the VOI is assigned, as shown in Figure 3-2. This is the reason that an average [] is needed in the calculation. 21 Figure 3-2. Illustration. An example of S~hn, the set of vehicles not causing a HN to a VOI We recall that the aforementioned calculations give a complete framework for obtaining the probability that a BSM experiences a collision with another BSM that was transmitted by another vehicle. We suggest that with formulation that [SYNC vehicles] + [HN vehicles], one can approximate as Pcol = 1 P[All the other vehicles remain silent] 1 - (1 - ) (15) The key assumption that upholds the equation is that all the vehicles, even including those outside of the VOI's carrier-sense range, go through the same type of backoff process, which, as such, yields a Ptx in the same manner. 3.2.4. PDR: Packet Delivery Rate With Pcol calculated as in the previous subsection, the analysis proceeds to calculate the BSM delivery rate (PDR). This metric defines the probability that a BSM (i) has made it through the 22 Figure 4-1. Illustration. Structure of the testbed implementing a DSRC Tx CSMA/CA process and (ii) has not run into any collision over the air. As such, this quantity can measure the probability of a successful BSM delivery to an arbitrary vehicle within the transmitter vehicle's communication range. = (1) (1 - ) (16) = (1 - ) It is significant to notice that the PDR is a metric that represents an entire network; hence, it needs to be calculated as a result of considering all the vehicles in the network. This is what distinguishes PDR from the previous metrics--i.e., Ptx and Pcol, which were defined at each vehicle. It means that building on the previous quantities, the PDR serves as a metric that illuminates the entire network's perspective. We believe that this analysis principle will also be aligned with the interest of GDOT as a mission organization for the state of Georgia. It must be desired that how much reliability is achieved among all the vehicles on the road as an entirety of a network. In view of this, the metrics for calculation of the performance are defined in such a way that they consider all the vehicles in a network. 23 4. Experiments This project built a testbed for initial testing on the performance of GDOT's CV infrastructure. The testbed is expected to enable GDOT to assess its CV infrastructure fast and efficiently, given any decision on 5.9 GHz band by the FCC. The current status of the testbed is completion of establishing a DSRC-based BSM transmitted built on a software-defined radio (SDR) built on a field programmable gate array (FPGA), which is illustrated in Figure 4-1. We report that a unit of the FPGA was purchased via this project. Example scenarios applying the produced testbed include the performance evaluation of GDOT's CV infrastructure according to the American Association of State Highway Transportation Officials (AASHTO)'s Signal, Phasing and Timing (SPaT) Challenge (CAMP 2017) operating in DSRC under interference from C-V2X and/or Wi-Fi. The core part of the experimental outcome of this project is a suite of computer simulations. The software aims at evaluating the performance of a DSRC system with a large number of vehicles (i.e., Nsta), which is a purpose that a testbed is not able to efficiently serve. As such, the results that will be presented in Section 5 are mainly produced from the use of this software. Figure 4-2 illustrates the structure of the simulation software that has been built as an outcome of this project. We report that a unit of software license was purchased via this project. One of the key features of this simulation software is the user's ability to select key parameters, which guarantees the applicability to a wide variety of scenarios. To elaborate, running this software starts from setting the input parameters. 24 First, the user is allowed to select between a suburban or urban geographical setting. We made two scenarios ready in the simulation: namely, (i) suburban highway and (ii) urban canyon. The key output that will be differentiated as a result of selecting suburban or urban is Ncs, denoting the number of other vehicles sensed by a VOI. The main rationale is that physical objects (i.e., buildings) act to cause "blockage" of signals in an urban canyon, while no signal is blocked in a suburban scenario. Technical details that are induced by this difference will be elaborated through Sections 5.2 and 5.3. With the geographical scenario set, the user proceeds to configure other parameters. The parameters range from (i) V2X-related parameters (including CW, transmission range, etc.) and (ii) road-related parameters (including road segment length, lane width, number of lanes, vehicle density, etc.). In particular, in an urban canyon scenario, the user will also be asked to configure the number of buildings as a part of the road-related parameters. Once the parameters are set by the user, the program calculates Ptx by numerically solving simultaneous equations on Pb and Ptx, based on the equations that were provided in Section 3.2.1. The Ptx that are obtained as such will be used to calculate Psync and Phn, which will be used to compute Pcol. Then, the program proceeds to calculation of PDR based on the Pcol. 25 (a) Flowchart (b) Structure Figure 4-2. Illustration. Flowchart and structure of the simulation software 26 5. Results Results that are presented in this section were produced from running computer simulations and testbed that were explained in Section 4. In order to display the results in various viewpoints, we present this section in the following manner. In Section 4.1, we first show a general twodimensional (2-D) scenario, which will be followed by urban canyon and suburban scenarios in the subsequent subsections. 5.1. General 2-D Setting Figure 4-3. Screenshot. Distribution of vehicles on a general 2-D geographical setting 27 By a general 2-D scenario, we refer to distribution of vehicles on a "square"-shaped 2-D geographic area. Figure 4-3 displays an example of such a general setting. This will give a general idea of the DSRC networking performance as a function of the number of vehicles, which is the key quantification in the performance evaluation. An example of this general setting is illustrated in Figure 4-3. Although the dimension of the geographic area was set to be 1,000 m by 1,000 m, the user is able to change the dimension as needed. Another significant parameter that the user can configure is the range of BSM transmission for a vehicle, which is displayed as a black circle around a "star"-marked vehicle (i.e., VOI). As shown in the figure, the transmission range was set 200 m for the particular example. This is a parameter that affects the performance of a DSRC system because it determines (i) the number of other vehicles that can receive a BSM that is broadcast by the VOI and (ii) the number of vehicles "interfering" with the VOI. This is being depicted in the figure as different colors--i.e., blue circles and red circles indicating vehicles within and outside of the transmission range of VOI, respectively. It is also noteworthy that the vehicles are distributed "uniformly" along with X and Y axes on the 2-D space. This distribution gives a Poisson point process where the number of vehicles counted within an arbitrary finite area follows a Poisson distribution. 28 Figure 4-4. Graph. Ptx versus Nsta according to CW in a general 2-D setting 5.1.1. Ptx We recall that Ptx denotes the probability that vehicle is able to transmit after making it through a backoff process. In other words, it is a quantity that is defined at each vehicle, which is also dependent on some key parameters defining a DSRC system. In essence, a beaconing period contains a very large number of slotsi.e., Lbcn is as large as 750 with 50 msec and 66.7 sec for a beaconing period and a slot time, respectively. The numbers Lbcn 750 and time length of 50 msec come from the fact that a CCH takes 50 msec of an entire 100-msec beaconing period, based on IEEE 1609.4 that defines the "alternation" between CCH and SCH (IEEE 2011). Obviously, the Ptx decreases with a network being packed with a large Nsta, which is shown in Figure 4-4. The reason is that Pb increases as Nsta increases, as has been shown in Section 3.2.1. 29 It makes less probable that each BSM is able to make it through a backoff process (see Figure 31), which results in a lower Ptx. One important thing to notice here is that Nsta does not specify the type of a system: an interfering vehicle can also be from another system, viz., Wi-Fi or C-V2X. This assumption makes it possible to extend the framework to inter-technology interference, which is the main issue in the current setting of the 5.9 GHz band. Also, we assume a network with "saturated" traffic, which means that the air interface is filled with BSMs at all times. Another significant observation from Figure 4-4 is that a smaller CW leads to a higher Ptx. This can be understood in the sense that a higher CW inevitably keeps a BSM within a backoff process for a longer time while other key parameters (e.g., the length of a time slot and the interval between two BSMs) are kept the same. 30 Figure 4-5. Graph. Pcol versus Nsta according to CW in a general 2-D setting 5.1.2. Pcol We remind that Pcol is a quantity that is defined for each BSM. The BSM can either be collided by another BSM that has been sent by (i) a vehicle that was synchronized with the VOI or (ii) a hidden node that was not recognized by the VOI. It is straightforward from Figure 4-5 that the Pcol decreases as a network is packed with a large Nsta. The reason can be explained based on the definition of Pcol that as was explained in Section 3.2.3: Pcol = 1 P[All the other vehicles remain silent]. From the formula, one can easily find that it is less possible that all the other vehicles remain silent as the Nsta increases. 31 Another key observation from Figure 4-5 is that a smaller CW leads to a higher Pcol. This is due to the fact that a higher CW yields a higher Ptx which lowers P[All the other vehicles remain silent]. This observation leads to an understanding that it may be an efficient tactic to increase the CW in a "crowded" DSRC network with too many vehicles. One interesting observation is an inflection point on a curve of Pcol for a large value of CW--i.e., CW 63 in Figure 4-5. It intuitively means that, although small, there is a region of Nsta in which more vehicles can yield a lower Pcol, which sounds counter-intuitive. The explanation goes back to the definition formula: Pcol = 1 P[All the other vehicles remain silent] 1 - (1 - ). There is a tradeoff between: (i) as Nsta increases, the chance of all vehicles remaining silent decreases; versus (ii) with a smaller Ptx, the chance of all vehicles remaining silent increases. We have found that the impact of (ii) is greater than that of (i) on Pcol, which results in the counter-intuitive phenomenon. 32 Figure 4-6. Graph. PDR versus Nsta according to CW in a general 2-D setting 5.1.3. PDR Recall from Section 3.2.4 that PDR is a quantity that is defined for representing an entire DSRC network. We emphasize that this metric will provide the GDOT engineers with a fair insight on a network as an entirety, rather than myopic sight on an individual vehicle. It is forthright from Figure 4-6 that the PDR decreases as a network is packed with a large Nsta. The reason is attributed the definition of PDR, which inherently is proportional to Ptx and inversely proportional to Pcol, respectively: i.e., PDR = (1 - ). As such, it is straightforward from the relationships that a Nsta yields a lower Ptx and a higher Pcol, which results in a lower PDR. 33 Another significant observation from Figure 4-6 is that a larger CW leads to a higher PDR. Recall that a larger CW resulted in a lower Ptx but a higher 1 - Pcol, two of the key components in the definition of a PDR. Therefore, the phenomenon implies that a DSRC network is driven more significantly by the "collision" of a BSM rather than the "delay" caused by a longer backoff. It also suggests that in a DSRC network with a delay-constrained setting, the PDR's tendency in terms of Ptx and Pcol could change. It is also noteworthy that the inflection point (which was discussed in Section 5.1.2) also appears on a curve of PDR. One can easily see that the reason of the phenomenon is attributed to the Pcol. 34 Figure 4-7. Screenshot. A snapshot of the urban canyon scenario (100 vehicles (small rectangles 2 m-by-7 m in size) and 200 buildings (grey squares, 50 m-by-50 m in size) uniformly randomly distributed on a 6-lane, 4-way junction in a 1 km-by-1 km two-dimensional space) 35 5.2. Urban Canyon Scenario As an effort to apply the findings to realistic settings, we created an urban canyon scenario and ran the simulations in it. 5.2.1. Key Performance Determining Factor The key factor affecting the DSRC performance in such an unban setting is the "blockage" of a BSM by physical objects on the side of the road. That is, in a city road, the vehicle traffic becomes denser, which leads to (i) a lower mobility of the vehicles and (ii) a higher chance of non-line-of-sight (NLOS) signal propagations. In particular, the blockage limits not only the communications range of a vehicle but the physical distribution of Rx vehicles as well. Table 4-1 depicts how the blockage changes according to the vehicle density. Notice that the vehicle density is defined as the number of vehicles within a 1 km-by-1 km 2-D space as illustrated in Figure 4-7. The "average number of vehicles within a transmission range" on the second column of Table 4-1 gives the average number of potential Rx vehicles within the communications range of a Tx vehicle. The third column indicates the average number of vehicles whose propagations are "blocked" by a building. The buildings are expressed as a grey rectangle in Figure 4-7. The last column of Table 4-1 gives the ratio of blockage, which is defined as # blocked BSMs Blockage ratio = Total # BSMs broadcast by a Tx vehicle A "blocked" and "successfully received" BSM is indicated as a red and blue beam, respectively, in Figure 4-7. 36 Table 4-1. Blockage rate according to vehicle density Vehicle density Average number of Average number of Blockage ratio (# vehicles/km2) vehicles within a vehicles blocked by transmission range buildings 50 27.7280 8.2464 0.2974 100 57.5030 16.6038 0.2887 150 86.6730 24.6323 0.2842 200 115.9570 32.8680 0.2835 250 145.7280 41.2863 0.2833 300 175.3120 49.8600 0.2844 5.2.2. DSRC Performance Taking into account the blockage caused by physical objects, we calculate the performance of a DSRC network (in terms of PDR) as shown in Figure 4-9. The figure shows that with CW of 15 and 127, a vehicle can achieve a higher PDR in an urban scenario. The rationale is that that the blockage acts to reduce the number of neighboring vehicles that "compete" for a channel. In other words, in an urban scenario, the receivability among the Rx vehicles not undergoing blockage is higher compared to a suburban setting. The reason is that the blockage also serves as physically dividing a large network into smaller ones. However, we would not recommend one taking the result in Figure 4-9 that an urban setting is more advantageous in terms of the DSRC BSM broadcast performance. This is because the PDR is not displaying the number of vehicles that received the BSM. In other words, the physical 37 "coverage" of a BSM broadcast must be suppressed in an urban setting compared to a suburban scenario where no blockage exists. Therefore, the third bar ("orange" in color) in Figure 4-9 shows the PDR that is "discounted" by the blockage rate, which gives the number of vehicles that are actually able to receive a BSM after consideration of the blockage. 38 Figure 4-8. Screenshot. A snapshot of the suburban highway scenario (100 vehicles uniformly randomly distributed on a 4-lane, 4-way junction in a 1 km-by-1 km two-dimensional space) 39 Figure 4-9. Graph. Comparison of DSRC performance (in terms of PDR) between urban and suburban scenarios (with CW = {31, 127}) 40 5.3. Suburban Highway Scenario As an effort to compare the findings from Section 5.2, we (i) created a suburban highway scenario, which provides a significantly different V2X communications environment from an urban canyon and (ii) ran the simulations in the suburban scenario. An example of such a suburban highway scenario is shown in Figure 4-8. The figure shows 100 vehicles distributed on a 4-way junction in a 1 km-by-1 km two-dimensional space. 5.3.1. Key Performance Determining Factor The key factor affecting the DSRC performance in a suburban setting is the "inter-vehicle interference." That is, in a suburban highway road, there are only few physical objects blocking the signals over the air (e.g., buildings). This acts to leave all the vehicles on an open space, which provides an easy environment to exchange signals, but, at the same time, an environment that is vulnerable to interference. 5.3.2. DSRC Performance In Figure 4-9, the higher interference appears to cause a lower PDR in a suburban setting as compared to an urban canyon. However, as we have discussed in Section 5.2.2, the outperformance of an urban scenario should be discounted by the blockage ratio since the higher PDR was achieved among a smaller number of vehicles. Through the comparison of the performance achieved from a suburban scenario to that from a "discounted" urban scenario, the suburban scenario is shown to: 1. Yield a higher PDR with a larger CW 41 2. Yield a higher margin of outperformance than a suburban operation with a smaller vehicle density In particular, the second point comes from the fact that an urban scenario is more fragile due to the blockage, which causes a higher blockage rate resulting in a larger discount in PDR. 42 6. Conclusions This project was devoted to investigating the performance of a DSRC network under a reduced amount of spectrum due to the latest decision by the federal government. The key outcomes of the project were (i) identification of performance metrics, (ii) establishment of a mathematical calculation framework, (iii) development of a suite of computer simulations, and (iv) collection of results in various traffic scenarios--viz., general, urban canyon, and suburban highway. The key findings can be summarized as: Probability of a BSM transmission is proportional to 1/CW Probability of an over-the-air BSM collision is proportional to 1/CW Probability of a successful BSM delivery is proportional to CW In an urban canyon scenario, the blockage caused by physical objects such as buildings is the main DSRC performance determining factor. In a suburban highway scenario, the inter-vehicle interference due to no blockage is the main DSRC performance determining factor. In terms of the probability of a successful BSM delivery, the DSRC performance in a suburban scenario is significantly limited with a smaller CW. The deliverables that are ready for submission to GDOT are identified as follows: A technical report describing the mathematical findings The simulations software suite Ex-parte letters to the FCC summarizing the findings A list of published articles based on the findings from this project follows: S. Kim, B. J. Kim, and B. Park, "Environment-adaptive multiple access for distributed V2X network: A reinforcement learning framework," in Proc. IEEE Vehicular 43 Technology Conference (VTC) 2021 Spring, To appear. S. Kim and B. J. Kim, "Crash risk-based prioritization of basic safety message in DSRC," IEEE Access, vol. 8, Nov. 2020. S. Kim and B. J. Kim, "Reinforcement learning for accident risk-adaptive V2X networking," in Proc. IEEE Vehicular Technology Conference (VTC) 2020 Fall. 44 References Almeida, T. T., L. de C. Gomes, F. M. Ortiz, Junior, J. G. R., and L. H. M. K. Costa. 2018. "IEEE 802.11p performance evaluation: simulations vs. real experiments." IEEE ISTC. Bianchi, G. 1998. "IEEE 802.11--Saturation throughput analysis." IEEE Communications Letters 12 (2). Bianchi, G. 2000. "Performance analysis of the IEEE 802.11 distributed coordination function." IEEE Journal on Selected Areas in Communications 18 (3). CAMP. 2017. SPaT challenge verification document. ver. 1.2. Cheng, B., H. Lu, A. Rostami, M. Gruteser, and J. B. Kenney. 2017. "Impact of 5.9 GHz spectrum sharing on DSRC performance." IEEE VNC. ElBatt, T., S. K. Goel, G. Holland, H. Krishnan, and J. Parikh. 2006. "Cooperative collision warning using dedicated short range wireless communications." ACM VANET. Fallah, Y. P., C. I. Huang, R. Sengupta, and H. Krishnan. 2010. "Analysis of information dissemination in vehicular ad-hoc networks with application to cooperative vehicle safety systems." IEEE Trans. Veh. Technol. 60 (1). Huang, C. I., Y. P. Fallah, R. Sengupta, and H. Krishnan. 2010. "Intervehicle transmission rate control for cooperative active safety system." IEEE Transactions on Intelligent Transportation Systems 12 (3). IEEE 802.11p Standard. Jun. 2010. 802.11p-2010 - IEEE Standard for Information technology Local and metropolitan area networks Specific requirements Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications Amendment 6: Wireless Access in Vehicular Envir. IEEE. IEEE. 2011. "IEEE 1609.4-2010, IEEE standard for wireless access in vehicular environments (WAVE)--multi-channel operation." IEEE. 2016. "IEEE 802.11-2016 - IEEE Standard for Information technology Telecommunications and information exchange between 14 systems Local and metropolitan area networksSpecific requirements - Part 11: Wireless LAN MAC and PHY Specifications." (IEEE). Jornod, G., A. El Assaad, A. Kwochzek, and T. Kurner. 2019. "Packet interreception time modeling for high-density platooning in varying surrounding traffic density." IEEE EuCNC. Kenney, J. Nov. 2018. "An update on V2X in the United States." SIP-adus Workshop on Connected and Automated Systems. 45 Khabazian, M, S Aissa, and M Mehmet-Ali. 2013. "Performance modeling of safety messages broadcast in vehicular ad hoc networks." IEEE Transactions on Intelligent Transportation Systems 14 (1). Khan, I., G. M. Hoang, and J. Harri. 2017. "Rethinking cooperative awareness for future V2X safety-critical applications." IEEE VNC. Kim, S. 2017. Coexistence of wireless systems for spectrum sharing. Blacksburg: Ph.D. Disertation, Virginia Tech. Kim, S, and B.-J. Kim. 2020. "Reinforcement learning for accident riskadaptive V2X networking." arXiv:2004.02379. Kim, S, and M Bennis. 2019. "Spatiotemporal analysis on broadcast performance of DSRC with external interference in 5.9 GHz band." arxiv:192.02537. https://arxiv.org/pdf/1912.02537.pdf. Kim, S, and T Dessalgn. 2019. "Danger aware vehicular networking." IEEE SoutheastCon. --. 2019. "Mitigation of civilian-to-military interference in DSRC for urban operations." IEEE MILCOM. Kloiber, B., J. Harri, T. Strang, S. Sand, and C. R. Garcia. 2015. "Random transmit power control for DSRC and its application to cooperative safety." IEEE Trans. Dependable Secur. Comput. 13 (1). Lei, X., and S. H. Rhee. 2019. "Performance analysis and enhancement of IEEE 802.11p beaconing." EURASIP Journal on Wireless Communications and Networking 2019 (61). Liu, Z., Z. Liu, Z. Meng, X. Yang, L. Pu, and L. Zhang. 2016. "Implementation and performance measurement of a V2X communication system for vehicle and pedestrian safety." Internat. J. Distributed Sensor Netw. 12 (9). NHTSA. 2016. "Proposed rule would mandate vehivle-to-vehicle (V2V) communication on light vehicles, allowing cars to 'talk' to each other to avoid crashes, NHTSA-34-16." December. https://one.nhtsa.gov/About-NHTSA/PressReleases/ci.nhtsa_v2v_proposed_rule_12132016.print. Qiu, Harry, Ivan Ho, Chi Tse, and Yu Xie. 2014. "Technical report: A methodology for studying 802.11p VANET." arXiv:1410.3978. https://arxiv.org/ftp/arxiv/papers/1410/1410.3978.pdf. Rendaa, M. E., G. Restaa, P. Santi, F. Martelli, and A. Franchini. 2016. "IEEE 802.11p VANets: experimental evaluation of packet inter-reception time." Elsevier Comput. Commun. 75. Son, S., and K.-J. Park. 2019. "BEAT: beacon inter-reception time ensured adaptive transmission for vehicle-to-vehicle safety communication." MDPI Sensors 19 (14). Stanica, R, E Chaput, and A Beylot. 2014. "Reverse back-off mechanism for safety vehicular ad hoc networks." Elsevier Ad Hoc Networks 16. 46 Torrent-Moreno, M., J. Mittag, P. Santi, and H. Hartenstein. 2009. "Vehicle-tovehicle communication: Fair transmit power control for safety-critical information." IEEE Transactions on Vehicular Technology 58 (7). US Department of Transportation. 2017. "Vehicle-to-vehicle communication technology, V2X Fact Sheet." July. Available: https://www.nhtsa.gov/sites/nhtsa.dot.gov/files/documents/v2v fact sheet 101414 v2a.pdf. US FCC. 2019. "Dedicated short range communications (DSRC) service." April. Available: https://www.fcc.gov/wireless/bureaudivisions/mobility-division/dedicated-short-rangecommunicationsdsrc-service. US FCC. Dec. 2019. In the matter of use of the 5.850-5.925 GHz band. Notice of Proposed Rulemaking, ET Docket No. 19-138. Wang, X, S Mao, and M. X. Gong. 2017. "An overview of 3GPP cellular vehicle-to-everything standards." GetMobile: Mobile Computing and Communications 21 (3): 19-25. Yao, Y, L Rao, and X Liu. 2013. "Performance and reliability analysis of IEEE 802.11p safety communication in a highway environment." IEEE Transactions on Vehicular Technology 62 (9). Yin, X, X Ma, and K. S. Trivedi. May 2013. "An interacting stochastic models approach for the performance evaluation of DSRC vehicular safety communication,." IEEE Transactions on Computers 62 (5). 47