Capacity Region of the Asynchronous Gaussian Vector Multiple-Access Channel

In this paper, we derive explicit expressions for the capacity region of the two-user symbol-asynchronous Gaussian vector multiple-access channel. Verdú considered the case where each user linearly modulates a fixed waveform in each symbol period, where the symbol periods for the users are not perfectly aligned at the receiver. He derived explicit capacity region expressions for the case where the transmitters have knowledge of the mutual offset and also for the case where the transmitters have no knowledge of the mutual offset. In this paper, we extend Verdú's results to allow each user to linearly modulate a set of orthonormal waveforms, instead of a single waveform, in each symbol period and with no restrictions imposed on the waveforms. We consider group power constraints, which include individual sum power constraints, as orthonormal waveforms assigned to each user may come from different frequency bands with different power constraints. Similar to the case where each user is allowed to linearly modulate only a single waveform, our results hold regardless of whether or not the transmitters are frame synchronous. In addition, we present some results that are necessary to numerically compute the capacity region expressions with general purpose convex optimization algorithms. Next, we simplify the capacity region expression when there are only individual sum power constraints and when the transmitters know the mutual offset. We also prove a sufficient condition for a similar simplification to hold when the transmitters have no knowledge of the mutual offset. Finally, we consider a specialized algorithm to numerically compute the simplified capacity region expression when the transmitters know the mutual offset.