ACK-based adaptive backoff for random access protocols

Abstract

In contention-based networks, ALOHA and CSMA are the most popular random access protocols, wherein a station transmits packets after a certain delay according to a particular backoff policy, such as binary exponential backoff (BEB). When a contention-based network is heavily loaded, system performance, in terms of transmission success rate and throughput, degrades dramatically under random backoff policies without considering the backoff choices of the neighborhood. To alleviate this problem, we propose and verify in this paper a highly efficient backoff policy, namely "ACK-based adaptive backoff " (AAB), which takes into account the total number of successful packet transmissions (with ACKs) among neighboring stations in determining a station's subsequent backoff period before its next packet (re)transmission. By doing this, AAB can effectively reduce the overall packet collisions under heavy-loaded conditions, even with hidden-stations. On comparing with BEB and its modified policy, multiple increase linear decrease (MILD), AAB can achieve about 17% and 10% performance gains of transmission success rate under CSMA protocol in a network with 20 stations with new packet arrival rate of 0.9, and 43% and 16% performance gains of energy efficiency respectively when the former two protocols reach their saturated points.