A conjugate direction search algorithm for ML estimation of frequency offsets in OFDMA uplink

This paper deals with the maximum-likelihood (ML) estimation of carrier frequency offsets (CFO) in OFDMA uplink transmissions. A pilot preamble-based approach is proposed that makes use of the conjugate direction search algorithm. In many earlier works, the multidimensional optimization involved in this joint estimation problem is broken into a series of one-dimensional searches. All these methods calculate the frequency offsets of the different users one after the other. However in this work, the Powell's method of Conjugate Direction Search is made use of, which gives all estimates simultaneously regardless of the number of active users in the system. Its performance is compared with a computationally-efficient approximation of the Alternating Projection algorithm [8]. The method compares favourably in this regard and has the added advantage of flexibilty of application to any subcarrier assignment scheme in OFDMA systems. The performance advantages of this new method are substantiated through simulation studies.