Jittering Broadcast Transmissions in MANETs: Quantification and Implementation Strategies
Roy Friedman, David A. Hay, Gabriel Kliot · 2009
Abstract—Delaying the transmission time of messages for a short random period, a.k.a. jittering, is a known approach for preventing concurrent transmissions, and consequently collisions, in multiple access networks (e.g., RFC 5148). It is particularly useful for increasing the reliability of unacknowledged broadcast messages in the 802.11 protocol. Yet, transmission jittering comes with a price. It increases the average end-to-end latency and could potentially decrease the throughput of the network. This paper investigates the relation between the maximal jitter duration and the probability of successful transmissions. Specifically, we propose and compare three implementation strategies: inside the MAC protocol, in the IP layer, and a cross-layer implementation. The comparison establishes that our practical crosslayer implementation minimizes the suggested protocol changes to 802.11 MAC protocol, while incurring the smallest latency, thus being the most attractive one. In fact, under IEEE 802.11e extension, our implementation does not require any MAC change. Our results show that by carefully designing the system, a very small jitter value of hundreds of microseconds suffices to obtain a high transmission success probability, a network utilization factor of above 70%, and subsequently to ensure high flooding reliability. This is in contrast with prior works, which have suggested to use two orders of magnitude larger jitter (i.e., tens of milliseconds). All our results are backed up by extensive simulations. I.