Performance Improvement in Muli-user MIMO Networks via Interference Alignment

Abstract

Almost all wireless networks are interference limited. Interference management has been always a primary concern for large section of current wireless networks with exponentially growing devices, lack of centralized medium access, power management. Because of broadcast nature of the wireless channel, all signals from simultaneous transmissions from devices apart in the same space, are added to the desired signal at the receiver end. Therefore optimal spectrum efficiency in such systems mandates distributed, low complexity interference management strategies with very less overhead which should be far more superior than existing successive interference cancellation, highly complex multiuser detection techniques. In this thesis, a novel interference management scheme- “Interference alignment” scheme for multi user scenario is investigated and analysed supporting the arguments with numerical results for most scenarios. Firstly, the concept of interference channel, Degrees of Freedom were well established which are prerequisite in understanding the predicament of multi user wireless channels. Later on, interference alignment concept has been put forward stating its origin back from linear algebra. IA for K-user MIMO is studied. In a fully connected K-user network with perfect channel state information, IA minimizes the interference space dimension at intended receivers thus maximizing the achievable capacity of the entire channel and increasing the Multiplexing gain. Later on the idea of IA is extended to multi-hop networks. A practical cellular multi-hop wireless network is considered and distributed interference alignment technique is implemented which shows superior performance even in high interference case. All IA schemes assume that the channels are full rank richly scattered environments which in practise is not always possible. The idea of using relays to act as external scatters which increase the rank of effective channel observed is considered. So two novel distributed relaying schemes have been proposed modifying the existing IA scheme to fit the case for rank deficient channels and still achieve multiplexing gain on par with full rank channels. The proposed algorithms doesn’t require global channel state information at all nodes except at relay nodes, doesn’t need large symbol extensions, and still are able to enhance the sum capacity of the network