Partially coherent detection in rapidly time varying channels

We investigate the performance degradation of basic modulation schemes in a rapidly time varying channel using a first order auto regressive channel model. We propose a partially coherent detector for both noncoherent frequency shift keying (FSK) and differential phase shift keying (PSK) that exploits partial channel knowledge to enable the receiver to operate effectively in both fast and slow fading. The maximum likelihood rule (ML) obtained for the partially coherent FSK turns out to be a linear combination of coherent and noncoherent detection rule. Results demonstrate that significant performance improvement can be achieved over the best of coherent and noncoherent FSK detection in fast fading. We also propose a few adaptive schemes that vary the modulation scheme in response to degrading quality of the channel estimate between successive training symbols