The benefits of MIMO technology have made it a solution for the present and future wireless networking demands. Increasing the number of antennas is an intuitive approach for boosting the network capacity; however, processing load and implementation limitations put a practical bound on this goal. Recently a solution known as massive MIMO has shown that a very large antenna array at the base station can simplify the processing, in a way that even matched filter (MF) can be used for detection purpose.
The ultimate performance of massive MIMO can be achieved only under some optimistic assumptions about the channel and hardware deployment. In practice, there are some restrictions that do not allow the ultimate performance for a massive MIMO system. Under some realistic assumptions, an efficient use of all the resources becomes important in a way that the application of simple algorithms like MF and zero forcing (ZF) becomes questionable. Thus, in this thesis work, more efficient approach based on optimal minimum power beamforming is considered as the benchmark. The idea is to investigate the behavior of this algorithm and the performance differences with respect to some sub-optimal methods when the system dimensions grow large.
Two solutions for the minimum power beamforming are reviewed (SOCP and uplink-downlink duality). The solution that is on focus is based on the second order cone programming (SOCP). Intercell interference(ICI) plays a critical role in the SOCP algorithm as it couples the sub-problems at the base stations. Thus, a large dimension approximation for the optimal ICI, using random matrix theory tools, is derived which tackles both of the processing simplification and the backhaul exchange rate reduction goals. This approximation allows derivation of an approximated optimal intercell interference based on the channel statistics that results a procedure for decoupling the subproblems at base stations.
The comparison between the SOCP algorithm and the sub-optimal methods is carried out via simulation. The results show that the performance gap with respect to the sub-optimal methods grows when correlation between the antenna elements at the BS side increase. In a network simulation with 7 cell and 28 users, this gap remains significant even with 100 antennas at the BS side. These performance differences justify the application of more complex algorithms like SOCP for a MIMO system with a large antenna array when the practical restrictions are taken into account.
| Identifer | oai:union.ndltd.org:oulo.fi/oai:oulu.fi:nbnfioulu-201310161794 |
| Date | 21 October 2013 |
| Creators | Asgharimoghaddam, H. (Hossein) |
| Publisher | University of Oulu |
| Source Sets | University of Oulu |
| Language | English |
| Detected Language | English |
| Type | info:eu-repo/semantics/masterThesis, info:eu-repo/semantics/publishedVersion |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess, © Hossein Asgharimoghaddam, 2013 |
Page generated in 0.0049 seconds