Analysis of a Random Channel Access Scheme with Multi-Packet Reception

Abstract

A key advantage of viewing communications in wireless networks as multiple access rather than a plurality of point-to-point transmissions, is its robustness towards multiple access interference. Concurrent packet transmissions are allowed to coexist thus deviating from the traditional view of enforcing collision-footprints around the transmitter-receiver pairs. What are the performance gains of employing channel access strategy based on a multiple access channel in a multihop wireless network? We consider a wireless multihop network, where nodes have a joint decoding capability to resolve up to K multiple concurrent packet transmissions from other nodes in their range. The basic assumptions are that the packet transmissions are asynchronous, i.e., nodes are completely uncoordinated, and that the packet transmission at each node is based on a probabilistic model. In this paper, we show that a simple random access strategy for communication over such channels offers significant gains in throughput while reducing latency in congested wireless networks. More precisely, we characterize the throughput performance gains through an exact analysis for the case of K = 2 and also offer tight approximations for arbitrary K. Furthermore, we study the asymptotic throughput behavior and prove asymptotic optimality of random channel access over multiple access channel. I. INTRODUCTIONAd hoc wireless networking is poised to be the next revolution in data networks with plenty of applications ranging from surveillance and tracking to data mining and distributed computing. In such networks, transmission from a source to its destination is over multiple hops through other nodes that assist in relaying information. Communication among nodes is established without any infrastructure support in an uncoordinated fashion. Since packet transmissions from nodes are asynchronous and random over a common channel, efficient transmission protocols that achieve higher network performances are a necessity. Several communication protocols for random, uncoordinated channel access have ushered over the past decades. Examples of basic random access protocols include ALOHA, carrier sense multiple access (CSMA), token, and tree-based protocols.A fundamental assumption in the design of these protocols is that a node can decode single-user transmission successfully while concurrent transmissions interfere with each other resulting in decoding failure of overlapping packets. Efficiency of such access protocols varies from 18% offered by ALOHA [1] to 100% for TDMA or centralized scheduling methods 1 .

QoS provisioning with adaptive backoff algorithm for IEEE 802.11ac under multipacket reception

“…Nagaraj et al [9] provide an exact analysis for the throughput of pure Aloha for channels with MPR capability K = 2, and an approximate expression for K ≥ 2. Arun and Venkatesh, based on an order statistics scheme have derived throughput of pure Aloha for channels with arbitrary MPR capability [10].…”

Fair and efficient scheduling in wireless networks with successive interference cancellation

“…In the broader context of multi-packet reception, there exist several work on scheduling and MAC protocols [8]- [10]. Nevertheless, non of them are applicable to SIC-based networks.…”

“…While this model provides a useful abstraction for a general multi-packet reception network, it does not accurately capture the underlying mechanisms of SIC-based networks such as selection of the transmission rates. Other works on multi-packet reception, such as [9] and [10], assume even a simpler model in which a receiver is capable of receiving up to k packets simultaneously regardless of the transmission rates and channel conditions. The closest work to ours is due to Kumaran and Qian [11] where the authors consider the uplink scheduling problem with SIC.…”

On the complexity of wireless uplink scheduling with successive interference cancellation

“…In the broader context of multi-packet reception, there are several works on scheduling and MAC protocols [10]- [12], yet non of them is applicable to SIC-enabled networks. As an example, Zhao and Tong [10] developed a MAC protocol with multipacket reception capability considering a reception matrix as the underlying model.…”

“…While the reception matrix model is a clever abstraction of a general multi-packet reception network, it does not accurately model the underlying dynamics of a SIC-based network such as the selection of transmission rates or the order of decoding. Other works on multi-packet reception, such as [11] and [12], consider a simpler model in which a receiver is capable of receiving up to k simultaneous packets regardless of the transmission rates and channel conditions of users.…”