RBG matching: an innovative combinatorial approach for OFDMA resource allocation

OFDMA performs a fundamental role in wired/wireless communications. One of the key techniques in OFDMA is the resource allocation, which has been attaching much attention from both academia and industry. In this paper, we describe an innovative combinatorial method to study this problem. An OFDMA system will first be formulated into a random bipartite graph (RBG). To meet various system configurations and requirements, different matching methods will be proposed to perform subcarrier allocation. By studying the properties of RBG matching, we will obtain close-form formulas for outage probabilities so as to evaluate the performance of subcarrier allocation algorithms. It is then demonstrated that by exploiting the frequency diversity and multi-user diversity, the proposed matching method can minimize the outage probability with fairness assurance, and achieve the same diversity-multiplexing tradeoff as point-to-point OFDM systems. The induced subcarrier allocation algorithms also enjoy a sub-linear computation complexity of O(N2/3) for parallel implementations, where N is the number of subcarriers. Besides, the proposed RBG matching method only needs one-bit CSI feedback.