Fredrik Rusek

Robust UE Receiver with Interference Cancellation in LTE Advanced Heterogeneous Network

Heterogeneous network (HetNet) deployment is a key feature to increase the network capacity in LTE- Advanced and systems beyond. One major obstacle is the interference from neighbor base-stations which can significantly degrade the system performance. In this paper, we present downlink interference mitigation and cancellation algorithms for HetNets in the so-called almost blank subframe (ABS) scenario. In the ABS scenario, one must differentiate between two cases, namely (1) the non- colliding case: the common reference symbols (CRS) of the serving and the dominant interfering cell are not overlapped, and (2) the colliding case: the CRS of the serving cell and the dominant interfering cell are overlapped. For case (1) we propose a robust equalizer and a low-complexity variant of it. For case (2) we present the application of the space alternating generalized expectation-maximization (SAGE) with a maximum a-posteriori (MAP) criterion. Based on extensive realistic link-level simulations, the outcome for case (1) is that the proposed robust equalization technique is almost as good as the receiver with traditional CRS interference cancellation (IC) but with lower complexity and latency. For case (2), we find that the SAGE-MAP with three SAGE cycles/iterations is sufficient to achieve the same performance as the receiver with ideal CRS-IC.

A robust low-complexity MIMO detector for rank 4 LTE/LTE-A systems

This paper deals with MIMO detection for rank 4 3GPP Long-Term-Evolution (LTE) systems. The paper revolves around a previously known detector [1], which we shall refer to as RCSMLD (Reduced-Constellation-Size-Maximum-Likelihood-Detector). However, a direct application of the scheme in [1] to LTE/LTE-A rank 4 test cases results in unsatisfactory performance. The first contribution of the paper is to introduce several modifications that can jointly be applied to the basic RCSMLD scheme which, taken together, result in excellent performance. Our second contribution is the development of a highly efficient hardware structure for RCSMLD that allows for an implementation with very few multiplications.