On Robust Approaches to the Semidefinite Relaxation Detector

The semi definite relaxation detector is one of several approaches to quasi-maximum likelihood detection that have been proposed in the literature. This detector can provide near maximum likelihood performance at polynomial time complexity thereby sparking the interest of the research community. In this work, we investigate how the semi definite relaxation detector can be made robust to model/channel uncertainties. Several models of uncertainty are studied and some plausible techniques of making the semi definite relaxation detector robust to these uncertainties are described.

I. Introduction

Multiple input multiple output (MIMO) has been acclaimed as the key to achieving gigabit wireless communications [1]– [3]. While multiple antennas can provide a diversity gain, the renewed interest in MIMO systems stems from the spatial multiplexing gain. MIMO enables spatial multiplexing, thereby allowing multiple streams to be transmitted simultaneously without incurring additional penalty in terms of transmission power or bandwidth. In the context of multiuser MIMO, this spatial multiplexing manifests itself as space division multiple access (SDMA) whereby the different spatial signatures of the multiple users allow for simultaneous co-channel transmissions to these users with minimal interference to each other. More surprisingly, this spatial multiplexing gain can also be realized by multiple antennas co-located at a single transmitter receiver pair in the context of single user MIMO provided that the physical environment offers rich scattering. This allows the capacity of the MIMO system to scale linearly minimum number of antennas on either side of the link. With such a large gain in capacity offered by MIMO, it is of little surprise that MIMO has rapidly found its way into recent standards such as IEEE 802.11n.