On Channel Access Design for Wireless Networks with Multi-Packet Reception
Ke Li · ERA: Education and Research Archive (University of Alberta) · 2015
As wireless devices have emerged as a ubiquitous part of people's everyday lives, the demands for faster wireless communications become even more pressing. Fortunately, the advanced techniques of the physical layer such as multiple-input and multiple-output (MIMO), multi-user detection (MUD), advanced modulation, etc., make multi-packet transmission (MPT) and multi-packet reception (MPR) possible. It has been well recognized that the MPT/MPR technique can improve the performance of the wireless networks. However, novel algorithms at the medium-access control (MAC) and higher layers are needed to fully exploit the MPT/MPR capability. In this thesis, we study the behavior the MPT/MPR wireless network, evaluate its potential performance and design algorithms to efficiently and fairly manage the MPT/MPR networks. We start from a single-hop scenario where uncoordinated nodes share a MPR channel and assess its performance by designing additive-increase multiplicative-decrease MAC (AIMD-MAC) to achieve the max-min fairness. We show that with an appropriate set of parameters, AIMD-MAC can be applied to distributed environments where the number of nodes and channel capacity are not constant to achieve at least 90% of the performance of the benchmark. For multi-hop scenarios, we observe the M property of MPT/MPR networks, which profoundly changes the traditional understanding of managing a multi-hop wireless network. By identifying and investigating the M property, we propose novel algorithms to evaluate the MPT/MPR networks and demonstrate the relative importance of the MPT and MPR capacity limits. To efficiently manage the multi-hop flows traversing a MPT/MPR network, we design the AIMD backpressure MAC (AB-MAC) algorithm. Extensive simulations show that AB-MAC significantly outperforms IEEE 802.11 especially in dense networks.