An Outage Exponent Region based Coded f-Matching Framework for Channel Allocation in Multi-carrier Multi-access Channels

TL;DR

This paper introduces an analytic framework using outage exponent regions for channel allocation in multi-carrier multi-access systems, optimizing performance with a novel coded f-matching approach and low-CSI complexity algorithms.

Contribution

It proposes a new outage exponent region based performance metric and a coded f-matching framework with a low-CSI RBG approach for optimal channel allocation.

Findings

Achieves full frequency diversity and optimal outage exponent region.

Develops a low-complexity parallel algorithm for channel matching.

Provides closed-form formulas for outage probability and performance metrics.

Abstract

The multi-carrier multi-access technique is widely adopt in future wireless communication systems, such as IEEE 802.16m and 3GPP LTE-A. The channel resources allocation in multi-carrier multi-access channel, which can greatly improve the system throughput with QoS assurance, thus attracted much attention from both academia and industry. There lacks, however, an analytic framework with a comprehensive performance metric, such that it is difficult to fully exploit the potentials of channel allocation. This paper will propose an analytic coded fmatching framework, where the outage exponent region (OER) will be defined as the performance metric. The OER determines the relationship of the outage performance among all of the users in the full SNR range, and converges to the diversity-multiplexing region (DMR) when SNR tends to infinity. To achieve the optimal OER and DMR, the random bipartite graph (RBG) approach, only depending on 1 bit CSI, will be proposed to formulate this problem. Based on the RBG formulation, the optimal frequency-domain coding based maximum f-matching method is then proposed. By analyzing the combinatorial structure of the RBG based coded f-matching with the help of saddlepoint approximation, the outage probability of each user, OER, and DMR will be derived in closed-form formulas. It will be shown that all of the users share the total multiplexing gain according to their rate requirements, while achieving the full frequency diversity, i.e., the optimal OER and DMR. Based on the principle of parallel computations, the parallel vertices expansion & random rotation based Hopcroft-Karp (PVER2HK) algorithm, which enjoys a logarithmic polynomial complexity, will be proposed. The simulation results will not only verify the theoretical derivations, but also show the significant performance gains.