For two-user interference channels, the capacity is known for the case where interference is stronger than the desired signal. Moreover, it is known that if the interference is above a certain level, it does not reduce the capacity at all. To achieve this capacity, the channel inputs need to be Gaussian distributed. However, Gaussian signals are continuous and unbounded. Thus, they are not well suited for practical applications. In this paper, we investigate the achievable rates if the channel inputs are restricted to finite constellations. Moreover, we will show by numerical simulations that rotating one of these input alphabets in the complex plane can increase the achievable rate region. Finally, we show that the threshold at which the single-user rates are achieved also depends on this rotation.

Abstract-We consider a Gaussian multiple-access channel (MAC) with an amplify-and-forward (AF) relay, where all nodes except the receiver have multiple antennas and the direct links between transmitters and receivers are neglected. Thus, spatial processing can be applied both at the transmitters and at the relay, which is subject to optimization for increasing the data rates. In general, this optimization problem is non-convex and hard to solve. While in prior work on this problem, it is assumed that all transmitters access the channel jointly, we propose a solution where each transmitter accesses the channel exclusively, using a time-division multiple-access (TDMA) scheme. It is shown that this scheme provides higher achievable sum rates, which raises the question of the need for TDMA to achieve the general capacity region of MACs with AF relay.

No abstract

SUMMARYIn this paper, we consider a multiple-input-multiple-output-orthogonal frequency division multiplexing (MIMO-OFDM) downlink scenario, where each receiving mobile station has quality of service requirements, namely minimum rate requirements. For this problem we propose three heuristic resource allocation algorithms, which have a much lower complexity than the existing optimal solution (opt). We compare and evaluate these algorithms according to sum rate performance and complexity. The first strategy is based on a heuristic sum rate maximisation algorithm using the so-called eigenvalue updates. In our second algorithm, we make use of the duality of uplink and downlink, which allows us to do the allocation in the dual uplink. Finally, our third algorithm is based on the well-known zero-forcing dirty paper coding (ZF-DPC) principles, which use the Gram-Schmidt process to orthogonalise the transmissions towards the different users.

No abstract