UC Berkeley

This thesis studies the joint design of control, estimation, and communication sys-tems, focusing on their application to active vehicle safety systems. The active safetysystems on-board a vehicle, such as collision avoidance systems, help the driver de-tect and defuse oncoming potential crash situations. They have received signi¯cantattention recently. Wireless communication and estimation are critical componentsof an active safety system enabling the vehicle to be aware of surrounding vehiclesand the roadway. The information provided by the communication and estimationsystems can be used by vehicle control systems such as Cooperative Adaptive CruiseControl (CACC) or driver assistance/warning systems.

This thesis makes advances in three areas | a bene¯t analysis of vehicle-vehiclecommunication-aided CACC system in highway tra±c, the design of Medium AccessControl (MAC) protocols for vehicle safety messaging over Dedicated Short RangeCommunication (DSRC), and the real-time estimation of a Markov process over anoisy digital communication channel.We compare the in°uence of adaptive cruise control (ACC) and CACC systems onhighway tra±c behaviors. The main di®erence between the CACC systems and theACC systems is that CACC systems have vehicles interact by exchanging informationthrough communication. Communication is shown to decrease braking e®ort, increaseaverage velocity, reduce congestion, and smooth shock waves. The positive resultsmotivate us to design e±cient communication and estimation systems for vehicleactive safety system.We next study the design of MAC protocols for vehicles to send safety messages toeach other. The target is to send vehicle safety messages with high reliability and lowdelay. We propose and analyze several random access MAC protocols. The protocolsare compatible with the national DSRC multi-channel architecture. Both analyti-cal and simulation results show our approach is feasible for vehicle safety messagesin DSRC. Our best protocol outperforms the popular standard IEEE 802.11 MACprotocol.Finally, we study the real-time estimation of a Markov process over a memorylessnoisy digital communication channel. The problem could model a vehicle estimating3in real time the position, relative velocity, and direction of neighboring vehicles usingthe information obtained by wireless communication. We derive structural resultson the optimal encoder and decoder are derived. A recursive algorithm is given tojointly ¯nd a locally optimal encoder and decoder for the binary symmetric channel.For a memoryless Gaussian vector source and a binary symmetric channel, we ¯ndthe optimal policy and derive the minimum mean squared error as a function of thevariance of the source and the channel noise.