The Efficiency and Delay of Distributed Source Coding in Random Access Sensor Networks

Abstract

The reliability and delay of Slepian-Wolf distributed source coding (DSC) in sensor networks is analyzed under the random access setting. Consider a network of N sensors that observes correlated information from the environment and sends the local data to a central processor through direct transmission links. Due to the low message rate in sensor networks, we adopt the slotted ALOHA random access protocol where the time is divided into synchronized time slots and each sensor is allowed to access the time slots with independent probabilities. To eliminate the redundancy in the transmitted data, the sensors encode the local messages based on the Slepian-Wolf DSC method. Specifically, we assume that the sensors' message are encoded with a sequential dependency among each other and, thus, must be decoded one after the other such that the decoding of a particular message is reliant on the successful decoding of all the messages encoded earlier in the sequence. In this case, the loss of one message may result in the failure of other messages and the delay in the successful decoding of a particular message also varies from sensor to sensor. In this work, we analyze the performance of Slepian-Wolf DSC in random access networks in terms of the rate of successful decoding and the average delay of each message. Specifically, we propose and compare different transmission probability assignments for DSC in the ALOHA network and emphasize the importance of the MAC design.