Broadcast Collision Mitigation in Vehicular Ad Hoc Networks

Whereas passive safety systems more and more reach their physical limits, active safety systems are going to determine the road map towards a "zero-accidents" vision. A big potential is seen in Vehicular Adhoc NETworks (VANETs). They are expected to go far beyond the capabilities of local radar- and vision-based sensors, by providing an enhanced view of the current environment, known as cooperative awareness. Therefore, vehicles are compelled to periodically broadcast safety-related information (e.g. position, speed, heading) to their neighbors. For that they use ITS-G5, a fully decentralized communications technology based on IEEE 802.11. IEEE 802.11's PHYsical (PHY) and Medium Access Control (MAC) layer have been slightly adapted to support vehicular environments (e.g. multi-path delays, transient connections). However, VANETs introduce new cooperative safety applications, di�ering significantly in terms of communications policies and requirements. Especially the broadcast policy for safety-related information disables IEEE 802.11's collision avoidance mechanisms, like the exponential increase of the contention window, or the Request To Send/Clear To Send (RTS/CTS) handshake. Without them, the amount of packet collisions increases signi�cantly under heavy communication loads, resulting in a distinct degradation of the communications performance. To support new cooperative safety applications, providing a high awareness quality is life critical. This issue raises the following research question: How to transmit safety-related information with su�cient reliability by using a potentially undependable communications technology? From a MAC perspective, packet collisions are the main reason for undependable communications in VANETs. Hence, the �rst part of this work analyzes the sources of packet collisions, as well as their behavior in space and time. The results show, for instance, that a significant amount of packet collisions at close ranges are caused by vehicles, having chosen the same backo� counter (waiting time). Moreover, subsequent packet collisions reveal a (temporal) correlated behavior that may significantly degrade safety-related measures like the update delay or inter-reception time. Based on these results, three new broadcast collision mitigation strategies are introduced and evaluated in the second part: First, the geo-backo� concept aims at relocating packet collisions from near to far, as close ranges are much more critical than farther ones for vehicular safety. Therefore, position information is exploited in order to generate the backoffcounter. Second, the random transmit jitter concept addresses correlated packet collisions in the time domain, by randomizing the periodic safety-broadcasts around the nominal transmission interval. Finally, the concept of random transmit powers alleviates the same issue by randomizing collision and interference areas in space. From an applications perspective, the important question is more about whether applications requirements are ful�lled, and not about the how. Hence, in the �nal part the impact of the previously introduced broadcast collision mitigation strategies on applicationspecific requirements is nvestigated. Therefore, a new awareness control strategy is proposed. It implements a framework called �sh-eye awareness, speci�cally, it allows to adapt the awareness quality as a function of the range. In a �rst step, Random Transmit Power Control (RTPC) is proposed. It manages to provide di�erent levels of awareness quality at different ranges, while mitigating correlated packet collisions by randomizing them in space. Because RTPC is able to reduce the channel load, the second step is to combine RTPC with Transmit Rate Control (TRC), and bene�t from the gained channel resources by subsequently increasing the transmit rate, and by implication, the quality of the awareness. The Fish-eye Awareness Control (FAC) strategy is evaluated through simulations, with focus on cooperative driving applications, such as platooning. Finally, the geo-backoff and random transmit jitter are integrated as well, to bene�t from these collision mitigation strategies in addition