Performance Analysis of a Coherent FH-MA System Based on Latin Squares

Frequency hopping (FH) combined with coding is a popular technique in wireless communication applications. Motivated by the commercial system outlined in [1], this paper investigates the performance of a coherent FH multiple access (FH-MA) cellular system, where hopping patterns are constructed from latin squares. Some of the patterns carry pilot symbols, which the users track to estimate the channel on the data patterns and perform coherent detection. The latin squares construction enables us to analytically evaluate the bit error probability (BEP) over each data pattern, when convolutional codes are employed for error protection. The focus of our work is to precisely determine the effect that frequency and interference diversity, ensuing from FH, have on the coded performance. Regarding frequency diversity, it is shown that the performance can vary widely for different latin squares. Moreover, under pilot-assisted channel estimation, a performance tradeoff is observed, as an increasing frequency selectivity results in more diversity, but also more inaccurate channel state information. With respect to interference diversity, we analyze the performance when the interference power variations are perfectly known or ignored by the receiver. Our results confirm that interference tracking is important in harnessing the error correcting power of convolutional coding.