Performance of orthogonal short-time Fourier signaling over doubly dispersive channels

Orthogonal short-time Fourier (STF) basis functions are natural for communication over channels that are selective in both time and frequency. However, the STF basis functions in general interfere with each other due to the loss of orthogonality by channel dispersion. It is shown that the channel spread factor, the product of multipath and Doppler spreads, plays a key role in determining system performance. Smaller spread factors result in lower interference. A set of appropriately chosen STF basis functions serve as approximate channel eigenfunctions. A simple and approximately optimal pulse scale adaptation rule is derived to minimize interference by matching pulse properties to channel characteristics. It is also shown that ergodic capacity of doubly dispersive channels deviates from that of flat fading channels as channel spread factor increases. In particular, moments of the eigenvalue distribution of dispersive channels agree with those of flat fading channels up to certain order that is inversely proportional to channel spread factor.