Reliable Broadcast of Safety Messages in Vehicular Ad hoc Networks. Farzad Hassanzadeh

Transcription

1 Reliable Broadcast of Safety Messages in Vehicular Ad hoc Networks by Farzad Hassanzadeh A thesis submitted in conformity with the requirements for the degree of Master of Applied Science Graduate Department of Electrical and Computer Engineering University of Toronto Copyright c 2008 by Farzad Hassanzadeh

2 Abstract Reliable Broadcast of Safety Messages in Vehicular Ad hoc Networks Farzad Hassanzadeh Master of Applied Science Graduate Department of Electrical and Computer Engineering University of Toronto 2008 Broadcast communications is critically important in vehicular networks. Many safety applications need safety warning messages to be broadcast to all vehicles present in an area. In this thesis, we propose a novel repetition-based broadcast protocol based on optical orthogonal codes. Optical orthogonal codes are used because of their ability to reduce the possibility of collision. We present a detailed mathematical analysis for obtaining the probability of success and the average delay. Furthermore, we propose to use coding to increase network throughput, and adaptive elimination of potentially colliding transmissions to further increase reliability. We show, by analysis and simulations, that the proposed protocol outperforms existing repetition-based ones and provides reliable broadcast communications and can reliably deliver safety messages under load conditions deemed to be common in vehicular environments. We also show that the proposed protocol is able to provide different levels of quality of service. ii

3 Dedication To my beloved sister, Farnaz iii

4 Acknowledgements I wish to express my sincerest thanks to my supervisor, Prof. Shahrokh Valaee, whose support made this work possible. His knowledge, guidance, and encouragement has been of invaluable help to me in producing this work. I would also like to thank my colleagues at Wireless and Internet Research Laboratory (WIRLab) for their constructive comments. This work was supported by AUTO21 Network of Centres of Excellence and in-kind contributions of Mark IV Industries. iv

5 Contents 1 Introduction Motivation and Purpose Infrastructure Networks and Ad Hoc Networks Collision Avoidance Operation: A Map of the Neighbourhood Delay Requirement of Vehicular Communication Overview of Vehicular Communications Standards Scope and Objectives Contributions Background CSMA/CA-based Broadcast Protocols IEEE Wireless LAN Broadcast Support Multiple Access (BSMA) Broadcast Medium Window (BMW) Batch Mode Multicast MAC (BMMM) Location Aware Multicast MAC (LAMM) TRAcking DEtection (TRADE) Distance Defer Transfer (DDT) Urban Multihop Broadcast (UMB) Reliable Reservation ALOHA (RR-ALOHA) v

6 2.3 Repetition-based Broadcast Protocols Synchronous p-persistent Repetition (SPR) Synchronous Fixed Repetition (SFR) Other Protocols and Chapter Summary OOC-based Broadcast Protocol Broadcast using Optical Orthogonal Codes CDMA, Positive Optical Systems, and OOC Distributed Code Assignment Code Information Response Window Frame-Synchronous and Frame-Asynchronous OOC Code Generation Analytical Performance Study Probability of Success Probability of Success for SPR Probability of Success for SFR and OOC Probability of Success and Interference Probabilities Average Delay SPR SFR and OOC Numerical Results Adaptive Elimination, Coding, and QoS Provisioning Adaptive Elimination Coding Probability of Success with Coding Quality of Service Appendix: Proof of Proposition vi

7 6 Simulation Results Channel Model Protocol Performance Simulation Setup Probability of Success Performance versus Distance Setup Results Discussion Conclusions and Future Work Future Work Bibliography 97 vii

8 List of Tables 4.1 Possible codewords and their probabilities for one interfering user Sample values of p 1 obtained by generating OOC codes for various L and w 56 viii

9 List of Figures 1.1 Fatalities per 100 million vehicle-miles US Department of Transportation ITS architecture Obstacle causes abnormal road situation Chain collision avoidance Intersection collision avoidance Broadcast ad hoc medium access control protocols BMW protocol under ideal conditions for transmitting one data packet BMMM protocol under ideal conditions for transmitting one data packet Blocking: N7 becomes blocked because N6 is not able to transmit CTS Far nodes are in yield state, only node B receives the broadcast message Division of time into frames and timeslots for repetition-based broadcast A transmission pattern and its binary representation CDMA, OOC-based optical CDMA, and, OOC-based broadcast Network association phase CIQ and CIR transmissions Timeline of CIQ and CIR transmissions of Fig Circular and linear representation of a transmission pattern Frame-synchronous and frame-asynchronous transmissions Distribution of 1 s in a sample OOC ix

10 4.1 Approximate and sample values of p 1 for OOC Probability of failure of SPR Optimum average number of transmissions, w, for SPR Probability of failure of SFR Probability of failure of OOC Optimum number of transmissions, w, for SPR and OOC Optimum probability of failure for OOC, SFR, and, SPR Delay of successful transmissions for OOC, SFR, and, SPR Both u 1 and u 2 are scheduled to transmit in the third timeslot u 2 disables its transmission in the third timeslot Probability of success of SPR for different r s, L = 64, N = 31, w = Probability of success of SFR for different r s, L = 64, N = 31, w = Probability of success of OOC for different r s, L = 64, N = 31, w = High and low priority messages with different number of transmissions Map of roadway and cars Probability of failure versus w, for µ p = Probability of failure for r = 2 versus w, for µ p = Probability of failure for r = 3 versus w, for µ p = Average delay versus w, for µ p = Probability failure versus average delay Probability of success versus average load, for w = Throughput versus average load, for w = Providing different QoS levels with OOC Effect of Adaptive Elimination Comparison of analytical and simulation results: P s Comparison of analytical and simulation results: P s with coding x

11 6.13 Comparison of analytical and simulation results: Delay Probability of success vs. distance from the receiver for L = Probability of success vs. distance from the receiver for L = Probability of success vs. distance from the receiver for different QoS levels Simulation with and without capture xi

12 Chapter 1 Introduction 1.1 Motivation and Purpose According to the World Health Organization (WHO), road accidents annually cause approximately 1.2 million deaths and 50 million injuries worldwide [1]. If preventive measures are not taken, traffic accident death is likely to become the third cause of the loss of disability-adjusted life years (DALY) 1 in 2020 from ninth place in 1990 [2]. However, fatalities caused by car crashes are, in principle, avoidable. 21,000 of the annual 43,000 road accident deaths in the US are caused by roadway departures and intersection related incidents [3]. This number can be significantly lowered by deploying local warning systems enabled by vehicular communications. Departing vehicles can inform other vehicles of their intention to exit the highway and arriving cars at intersections can send warning messages to other cars traversing that intersection. Studies show that in western Europe a mere 5km/hr decrease in average vehicle speeds could result in 25% decrease in deaths [1]. Policing speed limits will be notably easier and more efficient using wireless communication technologies. Governments and manufacturers have been increasingly investing to find new ways 1 DALYs are the sum of the years of life lost due to premature mortality and the years lost due to disability. 1

13 Chapter 1. Introduction 2 to improve the safety of drivers, occupants, and pedestrians. Until recently, these efforts have generally been following a passive and non-cooperative approach. A passive safety system tries to minimize the casualties and cost of a collision by using devices such as air bags and shock absorbers but it is not able to prevent collisions. Many manufacturers have been deploying non-cooperative digital technologies into their vehicles wherein each vehicle tries to reach maximum possible safety by making decisions based on information that it has obtained individually. These systems include advanced braking systems and cruise control systems. In the United States, from 1960 to 2005, the rate of fatalities per vehicle-miles decreased from 5.1 fatalities per 100 million vehicle-miles (MVM) 2 to 1.5 [4]. As observed from Fig. 1.1, in spite of a sharp decline form 1965 to 1995, this rate has been roughly constant from 2000 to This may indicate that, although passive/non-cooperative safety systems have been effective in decreasing fatalities, these approaches alone are no longer capable of significantly reducing fatalities beyond the current state. Therefore, we are urged to move from passive/non-cooperative safety systems to active/cooperative safety systems, which can prevent accidents. Active/cooperative safety systems are part of a broad range of emerging communications, electronics, and informatics technologies, unified under Intelligent Transportation Systems (ITS), being developed to fundamentally enhance safety and productivity in surface transportation. ITS technologies are designed to significantly improve road travel by preventing accidents, decreasing congestion and gridlock, and enhancing traffic management and enforcement. ITS development relies, at its core, on a communication platform enabling fast and reliable communication in vehicular environments. Dedicated Short Range Communication (DSRC) standard, adopted by IEEE and ASTM International 3 (ASTM E [5]), provides the communication platform required by ITS [6]. The 2 Vehicle-miles: Miles of travel by all types of motor vehicles as determined by the states on the basis of actual traffic counts and established estimating procedures. 3 Originally known as the American Society for Testing and Materials

14 Chapter 1. Introduction Fatalities per 100 million vehicle miles Year Fig. 1.1: Fatalities per 100 million vehicle-miles importance of DSRC can be observed in ITS architecture illustrated in Fig. 1.2: DSRC is the enabling technology for supporting both vehicle to vehicle (V2V) and vehicle to roadside infrastructure (V2R) communication. To accommodate the need of ITS for communication infrastructure, in 1999, the U.S. Federal Communication Commission (FCC) allocated 75MHz bandwidth at 5.9GHz [8] to public and private vehicular communication applications based on DSRC. The 75MHz bandwidth is divided into seven 10MHz channels. Among the seven designated channels, one is assigned to V2V public safety communication (ch 172), one is assigned to intersection public safety (ch 184), four channels are assigned to public safety and/or private applications (ch 174, ch 176, ch 180, ch 182), and one channel is the control channel (ch 178) used mainly for broadcast traffic. Our goal in this work is to provide a Medium Access Control (MAC) protocol in ad hoc mode for broadcast communication. Such a MAC protocol must be able to reliably deliver safety-critical messages. Due to stringent delay requirements of safety traffic, transmission delay of a protocol designed for vehicular communication must be very low. Furthermore, a vehicular MAC must be capable of supporting mobility and effectively coordinating tens of sources of broadcast traffic.

15 Chapter 1. Introduction 4 Travelers Centers Remote Traveler Support Maintenance and Construction Management Traffic Management Emergency Management Toll Administration Commercial Vehicle Administration Personal Information Access Information Service Provider Emissions Management Transit Management Fleet and Freight Management Archived Data management Fixed-Point to Fixed-Point Communication Wide Area Wireless (Mobile) Communication Vehicles Maintenance and Construction Vehicle Transit Vehicle Commercial Vehicle Emergency Vehicle Vehicle to Vehicle Communication Vehicle Dedicated Short Range Communication Roadway Security Monitoring Toll Collection Parking Management Field Commercial Vehicle Check Fig. 1.2: US Department of Transportation ITS architecture [7] The rest of this chapter is organized as follows. In Section 1.2 we consider the choice between infrastructure and ad hoc communication for safety systems. Characteristics of the communication traffic of cooperative safety systems are presented in Section 1.3, noting that knowledge of a relative map of neighbouring vehicles is effective in collision prevention. Delay requirements of safety messages are described in Section 1.4. An overview of vehicular communication standards is given in Section 1.5. The scope and objectives of this work are presented in Section 1.6 and the main contributions of this thesis are listed in Section 1.7.

16 Chapter 1. Introduction Infrastructure Networks and Ad Hoc Networks One of the main questions that must be explored for designing a MAC protocol is the need for and the possibility of using infrastructure in the network. In this section, we identify situations in which a vehicular network requires infrastructure and situations that can be handled in ad hoc mode by considering some of the envisioned safety applications. Line-of-Sight Collision Prevention In this situation, all the vehicles, bikes, and pedestrians involved in a dangerous situation are within line-of-sight of others. This has two consequences. First, drivers usually identify the danger but in cases that result in accidents, recognition of dangerous situation is so late that preventive actions taken by drivers are not effective in preventing a collision. Therefore, safety systems must be capable of warning drivers noticeably faster than their own ability to identify a dangerous situation. Second, the possibility of line-of-sight communication alleviates the need for infrastructure. We provide two examples of lineof-sight crash prevention in the following. Obstacle Information Dissemination Fig. 1.3 shows a sample scenario in which part of a road is closed due to an obstacle. Although in some cases, such as constructions, (visual or radio) warning signals, set up in advance, inform the drivers to decelerate, in many situations, such as a piece of fallen freight being in the road, warning signals are absent. In the latter case, vehicular communication can be used to disseminate information about abnormal road conditions. In Fig. 1.3 car 2 transmits a warning message to all cars traveling behind it and within its communication range. It can be observed that an ad hoc communication system is capable of delivering the warning message and roadside units are not necessary.

17 Chapter 1. Introduction Obstacle Fig. 1.3: Obstacle causes abnormal road situation Preventing Chain Collisions A similar but more dangerous situation occurs when a car is forced to stop due to a sudden event such as abnormal behavior of another vehicle or an accident as illustrated in Fig Sudden deceleration in a highway may result in a collision or even a chain of collisions. To avoid rear-end collisions, drivers rely on brake lights of the car ahead of them to be able to stop in time. Needless to say, close distance, high speed, slow driver reaction, poor visibility, and/or poor road condition may, and occasionally do, prevent appropriate stopping and lead to accidents. Warning messages can be useful in informing drivers faster than they would recognize the danger without receiving warning messages. Warning messages can be particularly effective in avoiding chain collisions because drivers reaction usually depends on the physical reaction of the car immediately in front of them, i.e., a driver does not recognize the danger until the car immediately ahead brakes. Radio warning messages, on the other hand, are free from this limitation and, hence, can prevent chain collisions. Non-Line-of-Sight Collision Prevention An example of non-line-of-sight collision prevention is an intersection collision warning system. Intersection collisions constitute a major category of traffic collisions that are largely preventible. As illustrated in Fig. 1.5, however, line-of-sight communication is

18 Chapter 1. Introduction Fig. 1.4: Chain collision avoidance not usually possible especially in more dangerous cases in which visual recognition is difficult. In this situation, the warning message should be relayed from one vehicle to another one by an intermediate node. In cases with high traffic density, relaying can be done by another vehicle in the line-of-sight of the vehicles entering the intersection. However, in most cases, a relaying vehicle is not present and a roadside unit is needed to relay the warning message. Fig. 1.5: Intersection collision avoidance In situations that a roadside unit is not necessary, vehicular ad hoc network can provide effective and inexpensive networks to support ITS safety applications. Furthermore, vehicular ad hoc networks can provide communication with lower delay by delivering mes-

19 Chapter 1. Introduction 8 sages from vehicle to vehicle, eliminating the delay caused by transmitting messages to infrastructure and back to vehicles. Because vehicles moving in the same direction have lower speed relative to each other than to roadside units, problems caused by mobility are also alleviated in ad hoc networks. Nevertheless, in cases in which line-of-sight communication is not possible and one cannot rely on the presence of other vehicles to relay critical messages the presence of roadside units is necessary. From the above discussion, we conclude that a hybrid solution in which vehicular ad hoc networks provide the communication platforms in highways and rural areas, and roadside units are utilized in intersections and urban areas, is best suited to enable ITS safety message delivery. 1.3 Collision Avoidance Operation: A Map of the Neighbourhood In this section, we study the characteristics of the communication traffic in a vehicular network in a channel dedicated to safety communication. We assume that safety systems installed on each vehicle require a map of relative position of neighbouring vehicles. If positions of neighbouring vehicles are known to the safety systems, many collisions can be avoided. The distances to neighbouring vehicles being known, a safety system can appropriately warn the driver of potentially dangerous situations. The map can be used by the safety system to assist the driver in changing lane, entering and exiting highways and main roads, avoiding unsafe close distance to the vehicle immediately in front, and many other operations a driver needs to perform. If the velocity of the neighbours is also known, each vehicle can predict future positions and avoid possible collision-prone situations. Building a local map in each vehicle requires that: 1) each vehicle be able to discover its own absolute or relative position, and 2) vehicles be able to communicate position

20 Chapter 1. Introduction 9 information. Discovering the position of a vehicle can be done via GPS [9], radio ranging techniques [10], and/or, radar. Our focus in this thesis is on providing the second requirement, i.e., the design of a medium access protocol that is capable of delivering position information messages, as well as other data. At 100km/hr, a vehicle moves 6m (approximately the accuracy of GPS) in 216ms. Therefore, update frequency of approximately 5 messages/second guarantees accurate and up-to-date maps. However, if vehicles can accurately estimate their velocity, position prediction can be used to alleviate the need for update and lessen the number of update messages required. A well-designed medium access control layer must be capable of successfully delivering messages with said frequencies. Since vehicular update messages need to deliver limited information such as vehicle ID, message ID, position, velocity, road condition, warning, etc, the size of these messages is under a few hundred bytes. Location information, on the Earth s surface, in a spherical system with fixed r coordinate and 1 cm resolution can be delivered with log 2 (2π m/10 2 m) + log 2 (π m/10 2 m) = bits where m is the Earth s radius. Relative location information within 100m (in a 200m 200m square centered at the reference point) in a Cartesian system with 1 cm resolution can be delivered with 2 log 2 (200 m/10 2 m) = bits. Assuming each vehicle transmits its position in absolute form and its velocity and the positions and the velocities of vehicles immediately in front, behind, left, and right in relative to the absolute position, ( ) = 324 bits or 41 bytes need to be transmitted. Adding 2 bytes for the ID of each vehicle, in total 51 bytes is needed. If about 49 bytes are allocated for other uses, such as obstacle and its position, emergency car and its position, emergency braking, etc, the length of the safety message is about 100 bytes. Therefore, in vehicular communications, safety messages are short compared to data or multimedia messages. Assuming 30m separation between adjacent cars traveling in the same lane, in 300m of

21 Chapter 1. Introduction 10 one lane of a road, there are approximately 10 cars. If the radius of the neighbourhood for which vehicles construct a local map is considered to be 300m, depending on the number of lanes, tens of vehicles are within the communication range of a vehicle. Any protocol designed for vehicular communication must be able to support tens of nodes in the network. Whether or not a map is used for collision avoidance, any medium access control protocol must be scalable and flexible. A vehicular communication system in a crowded highway spans hundreds of kilometers and thousands of cars. Although communication is local and vehicles far from each other need not communicate for safety reasons, vehicles that are physically close to each other must always be able to communicate regardless of the logical structure of a protocol. 1.4 Delay Requirement of Vehicular Communication An automatic safety system is successful if it can recognize a dangerous situation before the driver of a vehicle does. For example, if the car immediately ahead suddenly stops, the driver needs to detect the brake lights, decide that the brakes should be applied, and move the appropriate muscles to apply the brakes. The mental processing time, i.e., the time from the moment an event occurs until the moment a decision is made, is between 500ms to 1.2s, depending on how unexpected the event is [11]. Noting that the warning message alerting a driver, itself needs to be processed, we conclude that communication delay must not exceed ms. This value is, henceforth, called the lifetime of a safety message. Note that the lifetime value assumed here is in agreement with 5 messages per second update frequency explained in the previous section.

22 Chapter 1. Introduction Overview of Vehicular Communications Standards IEEE a is adopted as the base MAC/PHY layer standard for DSRC [12]. The modification to a, to make it suitable for vehicular communications and capable of supporting ITS applications, is called IEEE p standard for Wireless Access in Vehicular Environments (WAVE) [13]. WAVE is based on testing and analyses of wireless communications in a mobile environment published in [5]. Draft 3.05 of p WAVE is the most recent version [14]. According to IEEE p, a compliant vehicular communication network supports both vehicular on-board units (OBU) and roadside units (RSU). An RSU is similar to a wireless LAN access point and can provide communications with infrastructure [13]. Also, if required, an RSU must be able to allocate channels to OBUs. There is a third type of communicating nodes called Public Safety OBU (PSOBU) which is a vehicle with capabilities of providing services normally offered by RSUs. These units are mainly utilized in police cars, fire trucks, and ambulances in emergency situations. DSRC provides several channels (seven 10 MHz channels in North America) for communications which are divided into two categories: a control channel and service channels. The control channel is reserved for broadcasting and coordinating communications which generally takes place in other channels. Although DSRC devices are allowed to switch to a service channel, they must continuously monitor the control channel. There is no scanning and association as in the conventional All such operations are done via a beacon sent by RSUs in the control channel. While OBUs and RSUs are allowed to broadcast messages in the control channels, only RSUs can send beacon messages. The complement of WAVE in higher layers is IEEE 1609 which is a family of standards dealing with issues such as management and security of the networks [15]: Resource Manager: This standard provides a resource manager for WAVE, allowing communication between remote applications and vehicles.

23 Chapter 1. Introduction Security Services for Applications and Management Messages Networking Services: This standard addresses network layer issues Multi-channel Operation: This standard deals with communications through multiple channels. 1.6 Scope and Objectives In this thesis, we propose and analyze a reliable medium access control protocol with low delay for supporting safety message broadcasting in a vehicular network. We aim at developing a protocol that is capable of satisfying low delay requirements of safety messages while being able to deliver messages with high reliability. A major difference between an ad hoc network in vehicular environment and a conventional ad hoc network is that in vehicular networks, as discussed earlier, traffic is of broadcast type; routine safety messages are issued from all vehicles several times per second and are intended for all their neighbours. Transmission of safety messages must be reliable and with very low delay. Conventional MAC protocols for ad hoc networks are not designed to handle broadcast traffic from many nodes in the network. For example, as explained in more detail in Chapter 2, in IEEE no mechanism exists to reduce the probability of collision for broadcast traffic. In IEEE , Request To Send (RTS) and Clear To Send (CTS) packets are transmitted before unicast communications to avoid collisions. It may seem straight forward to add RTS/CTS handshake to broadcast communications as well. However, in vehicular communication, the length of broadcast messages is short and comparable to that of RTS. Therefore, the probability of collision is not significantly lower for RTS packets. The short length of messages also contributes to inefficiency since the payload (safety message) is not significantly larger than the overhead (RTS+CTS). Furthermore, RTS/CTS handshake needs to be performed with more than one receiver to obtain the same reliability as that of unicast communication.

24 Chapter 1. Introduction 13 A solution to reliable delivery of safety broadcast messages is repetition-based broadcast. In repetition-based broadcast, the safety message is transmitted several times according to a transmission pattern. While repetition increases the probability of successful delivery compared to the ALOHA approach of IEEE , previously proposed transmission patterns are random and have no mechanism for avoiding collision among users. In this work, after reviewing the literature in Chapter 2, we propose a new protocol based on Optical Orthogonal Codes (OOC) to decrease collisions in Chapter 3. Optical orthogonal codes, originating from optical code division multiple access, have good properties which, as we show with analysis and simulations, decrease the probability of collision when used as transmission patterns in a repetition-based broadcast protocol. We present a code assignment mechanism that assigns a transmission pattern to each user which is unique within its two-hop neighbourhood. The generation of optical orthogonal codes is also discussed. A detailed analytical performance study is provided in Chapter 4, resulting in expressions for the probability of success and the average delay. The analysis is valid for previously proposed repetition-based broadcast protocols, OOC-based repetition broadcast, and other repetition-based broadcast protocols with fixed number of transmissions. Numerical results, obtained from the analysis, provide performance evaluation for different protocols and indicate that repetition broadcast based on OOC exhibits higher probability of success and similar delay compared to other repetition-based protocols. Following the analytical performance study, in Chapter 5, improvements to the protocol are presented. Adaptive elimination is introduced to further increase the reliability of the protocol by eliminating potential collisions. Furthermore, the use of coding in the proposed protocol is explained. Coding helps increase the throughput of the network. Since QoS provisioning is a desirable feature of vehicular networks, a method for providing different QoS levels is also introduced.

25 Chapter 1. Introduction 14 In Chapter 6, simulation results are presented to show the performance of the proposed protocol, in an environment with a traffic load similar to that required by the ITS safety applications, and compared to the performance of other repetition-based protocols. Simulation results demonstrate the superior performance of OOC-based broadcast and show that its performance is improved by adaptive elimination and coding. Finally, in Chapter 7, we end this work with our concluding remarks and provide directions for future work. 1.7 Contributions The main contribution of this work is the design of a novel medium access control protocol for broadcast communication in vehicular communication networks. The proposed protocol is suitable for transmission of short routine safety messages issued by vehicles to inform others of their position and other useful information. The protocol makes use of good correlation properties of optical orthogonal codes to avoid collisions. The following lists the contributions of this work, the chapter in which they were presented, and any publication reporting them. MAC Design based on OOC: Introduced and studied OOC-based repetition protocols for broadcast communications (Chapter 3 and [16], [17]) Performance Study: Developed a generalized analytical framework not only capable of analyzing OOC-based broadcast but also previously proposed repetitionbased broadcast protocols and any repetition-based broadcast protocol with a fixed number of repetitions (Chapter 4 and [16]) and presented simulation results (Chapter 6 and [16], [17]) Coding, QoS, and Adaptive Elimination: Introduced and studied methods to improve reliability, throughput, and quality of service provisioning in repetition-

26 Chapter 1. Introduction 15 based broadcast (Chapter 5 and [16], [17])

27 Chapter 2 Background In this chapter, we look at previous works focusing on the reliability of broadcast communication in ad hoc networks (vehicular or otherwise). Protocols applicable to vehicular communications can be divided into five categories as shown in Fig Among these categories we discuss protocols from the first three categories: Carrier Sense Multiple Access/Collision Avoidance (CSMA/CA)-based protocols, reservation-based protocols, and repetition-based protocols. CSMA/CA-based protocols are modifications of IEEE that extend the use of Collision Avoidance (CA) mechanism, which in IEEE is used only for unicast communications, to broadcast and are discussed, along with IEEE , in Section 2.1. These modifications incorporate RTS/CTS/ACK handshake into IEEE broadcast to force hidden terminals into yield state in order to avoid collisions. They may require the handshake to be performed with all receivers, a subset of receivers, or only one receiver. Reliable Reservation ALOHA (RR-ALOHA), a slot reservation protocol suggested for vehicular networks [18], is discussed in Section 2.2. In RR-ALOHA users reserve a timeslot in a frame for transmission. Synchronous p-persistent Retransmission (SPR) and Synchronous Fixed Retransmis- 16

28 Chapter 2. Background 17 Broadcast Ad hoc MAC protocols CSMA/CA-based protocols Reservationbased protocols Repetition-based protocols Cluster-based protocols BTMA-based protocols RR-ALOHA SPR SFR OOC IEEE BSMA BMW BMMM LAMM TRADE DDT UMB Fig. 2.1: Broadcast ad hoc medium access control protocols sion (SFR) [19], repetition-based protocols proposed to solve broadcast problems, are discussed in Section 2.3. The proposed protocol in this thesis is also a repetition-based protocol. Analytical and simulation results are presented for SPR and SFR along with the proposed protocol in later chapters. The main challenge in cluster-based protocols is not the medium access control protocol but the formation of stable clusters. Therefore, these protocols are not considered here. Also, Busy Tone Multiple Access (BTMA) protocols, such as [20], cannot be used in the single control channel of DSRC and, therefore, not discussed in this chapter. For each protocol described in this chapter, we discuss its suitability for broadcast in vehicular environments.

29 Chapter 2. Background CSMA/CA-based Broadcast Protocols IEEE Wireless LAN In IEEE , Request To Send (RTS)/ Clear To Send (CTS) handshake is the main means of combating the hidden terminal problem. However, broadcasting nodes do not generally use RTS/CTS packets [21]. The lack of RTS/CTS handshake for broadcast communication results in unreliable and inefficient communication. The situation is worse in vehicular environments when there are tens of nodes within single-hop communication range. For unicast communications, acknowledgement (ACK) packets are transmitted by receiving nodes to ensure recovery if transmission has not been successful. For broadcast communication, however, acknowledgement is not required in IEEE [22]. This increases the unreliability of transmission of safety broadcast messages. Lack of acknowledgements also increases the risk of network instability. Since the transmitting node cannot detect an unsuccessful transmission, it fails to change the size of its Contention Window (CW) [23]. Therefore, under heavy load, the size of contention window is not increased to reduce the probability of collisions and the possibility of instability is increased. Although WAVE is built upon wireless LAN, the requirements for reliable and low delay communication, especially broadcast, are not achieved by using the current standard without appropriate modification and enhancements as commonly observed in the literature [6], [8], [20], [24] [32]. Furthermore, unicasting such short messages as in vehicular safety communications (see Section 1.3), with approach is very inefficient. According to a model devised by Bianchi [33], the maximum bandwidth utilization of a with RTS/CTS handshake, at 54 Mb/s, with payload size of 100 bytes is less than 7% [8]. Multiple RTS/CTS handshakes, as proposed by some of the protocols discussed in this chapter, will further

30 Chapter 2. Background 19 decrease the efficiency Broadcast Support Multiple Access (BSMA) A simple extension to IEEE which adds RTS/CTS signaling to broadcast packets is presented in [34]. The protocol works as follows. After going through carrier sensing and a contention phase, the source node broadcasts an RTS packet and waits a certain time for a CTS, which, if received, triggers DATA transmission. With this approach, however, the collision of CTS packets is very likely. To solve the problem of collision of CTS packets, the authors suggest: Using a radio with capture capability so that one CTS packet is successfully received. In this case, it is assumed that one of the CTS packets is sufficiently stronger than others and can be received with a capture-enabled radio. This assumption, however, is unrealistic since CTS packets are likely to have similar received power. The source node s radio detects the busy period resulting from a collision as a CTS and starts transmitting if it detects a busy period in the channel after transmitting RTS. In this case the main purpose of CTS packets, which is to force hidden terminals to yield, is overlooked. Hidden terminals cannot distinguish a busy period resulted from colliding CTS packets from any other collision. In Broadcast Support Multiple Access (BSMA) [35], NAK signaling is added to the above protocol. The collision of control packets, however, remains unresolved Broadcast Medium Window (BMW) The fundamental idea behind Broadcast Medium Window (BMW) [27] is to increase the reliability of broadcast transmission by transmitting the broadcast packet to every neighbour using RTS/CTS/ACK transmissions in a round robin fashion. If a neighbour

31 Chapter 2. Background 20 has already received the packet, the data packet is not transmitted. All transmissions in BMW also follow CSMA/CA rules. In BMW, each node maintains a list of its neighbours. Consider a node wishing to broadcast a queue of packets to its neighbours. The broadcasting node chooses one of its neighbours as the current receiver and transmits an RTS to that node. The receiving node transmits a CTS in which it includes the sequence number of the packet up to which all packets are received successfully. Then, the broadcasting node transmits packets starting from the sequence number indicated in the CTS packet using DATA/ACK transmissions. Meanwhile, all other nodes, other than the current receiver, are also receiving the packets and store all successfully received packets. The packet transmission to the current receiver continues until a new packet, which has not previously been transmitted to any other node, is transmitted to the receiver. At this point the broadcasting node chooses another neighbour and continues this process until all packets are transmitted. BMW is only efficient when the transmission queue contains several packets. Therefore, it is only suitable for burst traffic, or delay insensitive traffic which can be shaped into bursts of packets. When traffic arrives in the form of single packets in the presence of n neighbours, in ideal channel with no collision, as illustrated in Fig. 2.2, there are n contention periods, n RTS transmissions, n CTS transmissions, n ACK transmissions, and one DATA transmission. High number of control packets and contention periods induce unacceptably high delays, which is increased, even more, in realistic channel conditions with collisions, with back-offs, and retransmissions. Since traffic arrival in vehicular environments is in form of single packets with low delay requirements (in the range of 100ms) and there are tens of nodes in the network, BMW is not suitable for vehicular communications.

32 Chapter 2. Background 21 Contention Phase RTS1 CTS1 CTS1 RTS1 DATA ACK1 Contention Phase RTS2 CTS2 CTS2 RTS2 ACK2... Contentio n Phase RTSn CTSn CTS2 RTSn ACKn Fig. 2.2: BMW protocol under ideal conditions for transmitting one data packet Contention Phase RTS1 CTS1 RTS2 CTS2 CTS2 RTS2... RTS1 CTS1 RTS2 CTS2... RTSn CTSn DATA RAK1... ACK1 RAK2 ACK2... RAKn ACKn Fig. 2.3: BMMM protocol under ideal conditions for transmitting one data packet Batch Mode Multicast MAC (BMMM) Contention Phase RTS1 CTS1 RTS2 CTS2... RTS2 CTS2 RTS1 CTS1 RTS2 CTS2... RTSn CTSn DATA RAK1... ACK1 Batch Mode Multicast MAC (BMMM) [31] is a protocol that decreases the number of contention periods of BMW. This is done by sequentially transmitting a RTS to all neighbours and receive a CTS. DATA is transmitted after receiving the last CTS packet. After transmitting the DATA packet, the broadcasting node sequentially requests acknowledgements from the receivers via Request for Acknowledgement (RAK) packets. The ideal operation of the protocol, where channel is perfect and none of the receiving nodes are in yield state, is illustrated in Fig By decreasing the number of contention periods from n to 1, BMMM decreases the delay. However, BMMM does not decrease the significant overhead produced by control RAK2 ACK2... RAKn ACKn packets. Furthermore, note that the yield duration indicated in CTS packets is significantly increased. For example, receiving node 1 in Fig. 2.3 forces all its neighbours to yield state for the whole duration of transmissions to all nodes while its neighbours need to be in yield state only for the duration of DATA and RAK1. This unnecessary increase in yield state duration results in increased inefficiency. Probability of blocking and blocking propagation [36], also, increases because of long yield states. Blocking is illustrated in Fig N7 becomes blocked due to the following events:

33 Chapter 2. Background 22 At t 1, N1 transmits an RTS to N2. N3, N4, and N5 yield upon reception. At t 2, N2 transmits a CTS to N1. N6 yields upon reception. At t 3, N7 transmits an RTS to N6. N8, N9, and N10 yield upon reception. At t 4, N7 does not receive a CTS and goes into exponential back-off. While transmission from N7 would not interfere with any ongoing transmissions, N7 is prohibited from transmission. Other nodes in yield state may also result in blocking of other nodes. Longer durations indicated in RTS and CTS packets cause longer yield states and, therefore, the probability that a node in yield state becomes the recipient of an RTS is increased and so is the probability of blocking of the node transmitting the RTS. N3 N9 t2:cts t2:cts t3:rts N4 N1 N2 N6 N7 N8 t1:rts N5 N10 Fig. 2.4: Blocking: N7 becomes blocked because N6 is not able to transmit CTS Another problem with long yield durations indicated in CTS/RTS arises from high mobility in vehicular environments. A node outside the range of a RTS/CTS packet may move into the transmission range, not having heard the RTS/CTS. This node may potentially start transmission that collides with ongoing transmissions. When the duration indicated in RTS/CTS is larger, the probability of topology change during that time is higher and, therefore, the probability of collision due to topology change is also higher.

34 Chapter 2. Background Location Aware Multicast MAC (LAMM) Location Aware Multicast MAC (LAMM) [31] improves the performance of BMMM by introducing cover sets. A cover set of the receivers of a broadcast message, transmitted by a certain transmitter, is a subset of receivers whose radio coverage, is the same as the radio coverage of all receivers. Therefore, if only nodes in the cover set send CTS packets, the effect is the same as when all receiving nodes send CTS. Furthermore, only nodes in the cover set receive RAK and transmit ACK. Determining the cover set requires position estimation which can be obtained via GPS. While LAMM decreases the number of control transmissions, the amount of decrease is not discussed in [31]. The duration indicated in CTS/RTS control packets is also still larger than that of BMW. It might be possible to modify LAMM to adapt it to vehicular environment and decrease its overhead and CTS durations significantly. Exploring this possibility requires further research. The next three protocols introduce methods of finding one receiver and perform handshake with only that node instead of all the recipients to decrease the overhead caused by control packets TRAcking DEtection (TRADE) TRAcking DEtection (TRADE) protocol introduced in [37], is a protocol which uses location information to find the farthest node in each direction to eliminate the broadcast storm problem [22] caused by flooding in multi-hop broadcast. However, this protocol can be modified to increase the reliability of single-hop broadcast compared to IEEE By modifying a simplified version of TRADE we obtain a protocol that works as follows. The broadcasting node chooses the farthest node within its communication range, and unicasts the broadcast packet to that node in each direction on a straight road using RTS/CTS/ACK handshakes. Approximating the positions of vehicles as being on a straight line, if the farthest vehicle receives the message without collision, so do all vehicles

35 Chapter 2. Background 24 between the source and the farthest vehicle. This approach, however, has drawbacks. TRADE assumes the positions of vehicles are known by transmitting GPS messages. This can be problematic since GPS messages also need to be broadcast routinely. Since collisions are not the sole reason of packet loss, an intermediate node may still lose the packet due to noise, fading, or a large vehicle blocking line-of-sight while the farthest node receives it successfully. A broadcast message is of much more importance to the nearest vehicle than it is to the farthest. Therefore, if the farthest node is in yield state, the message is not transmitted and its lifetime may expire, while the closet node, which is its more important target, would be able to receive it without interference Distance Defer Transfer (DDT) Distance Defer Transfer (DDT) [37] is proposed to eliminate the need for a priori position information in TRADE, which is one of its drawbacks. DDT, similar to TRADE, is designed to combat the broadcast storm problem. However, it can be modified to increase the reliability of broadcast communication. In this modification of a simplified DDT protocol, the source node broadcasts the RTS packet in which the position of the sender is included. Each of the receiving nodes start a timer whose value is a decreasing function of their distance from the source. When a node s timer expires, if no other node has transmitted a CTS, that node transmits a CTS. This approach solves the first problem mentioned above, but the second and third problems remain unresolved. Furthermore, the likely collision of CTS s (or re-broadcasts in the original protocol) when the distance of two nodes is roughly the same from the source is ignored.

36 Chapter 2. Background Urban Multihop Broadcast (UMB) The design of Urban Multihop Broadcast (UMB) protocol [38] is aimed at finding the farthest node without a priori neighbourhood information. This protocol, basically, works as follows. The broadcasting node sends a Request-To-Broadcast (RTB) packet in which a desired direction of transmission is indicated. Each receiving node in the correct direction, transmits a jamming signal (black-burst), duration of which is proportional to its distance from the source and is an integral number of Short InterFrame Spaces (SIFS). After transmitting black-burst, each node listens to the channel. If it is the last node transmitting black-burst, it is chosen as the farthest node and continues to transmit a Clear-To-Broadcast (CTB) packet. The broadcasting node, then, transmits the DATA packet and receives an ACK packet from the farthest node. It is possible for two or more nodes to transmit black-bursts with the same length. In this case, they all transmit CTB packets leading to collision. If the broadcasting node senses the channel to be busy but is unable to decode the transmission, it assumes that two or more CTB packets have collided. In this case, it issues another RTB packet. In responding to the second RTB, only nodes that have transmitted after receiving the previous RTB compete for the channel by transmitting black-bursts. This procedure is iterated until a node successfully transmits CTB or a certain number of iterations is reached in which case the nodes that have transmitted the last CTB packets enter a random resolution phase which, if necessary, is repeated a certain number of times. If random resolution is also unsuccessful, the broadcasting node restarts the transmission from the beginning. Starting the node selection from the beginning may happen RET max times, after which the broadcasting node enters the back-off mode. While UMB manages to decrease the overhead, it puts too much emphasize on finding the farthest node while, as mentioned before, in vehicular communications, the closest node is the most important receiver. By the time the contention between far nodes is resolved, the life time of the message may have expired.