Out of group interference aware base station cooperation
Base Station Cooperation (BSC) technology enables the networked multiple input and multiple output (MIMO) transmission by setting up cooperation betwe en geographically separated Base Station (BSs). Past work showed BSC very effective in interf erence suppression and overall communication performance improvement. A lot of these work s commonly ignore that a rea- sonably sized network will not be able to operate in only one B SC group. This is because the cooperation between a large number of base stations will incur heavy signalling overhead, substantial latency and high levels of signalling. For this reason, there has to be multiple BSC groups operating in parallel, which will however cause a new kind of interference between BSC groups. We call it OGI - Out of Group Interference. This thesis shows that OGI will severely degrade system performance if ignored and make BSC unable to provide significant gain. To solve the problem, as the first main part of the contribution, we proposed OGI-aware MMSE precoding technique, which is proved to be at least 50% more effective than the traditional BSC joint precoding. Another precoding algorithm we propose is NB (noise balancing) BD (block diagonalization), which suppress the OGI and as NSD (null space decomposition) to block-diagonalize the effective channel matrices after precoding. This method achieves twice as much performance as the original NSD, which is the best performed traditional BSC precoder in our evaluation. Maximum Eigen-mode Transmission (MET) is a powerful algorithm to improve the performance of a Multi-user MIMO system. As second main part of contribution, we propose a combined MET-NBBD algorithm which leads to 25% better performance than NBBD. The other innovation we have, which is related to MET-NBBD, is a combined transmitter receiver interference suppression structure for OGI-aware BSC. This structure enables the receivers to jointly suppress OGI with transmitters and improve the performance to 45% better than NBBD. We also design a series of Proportional Fairness (PF) schedulers, which finds the suitable UEs by greedy searching, using channel norm versus past average rate as criteria. As third main part of the contribution, we evaluate and bench mark the aforementioned algorithms according to 3GPP (The 3rd Generation Partnership Project) evaluation requirement of CoMP (Coordinated Multi-Point). For the limited feedback scheme, we implement the Grass-mannian codebook based feedback. Different from previous results in part one and two, as they are based on inter-eNB cooperation, the simulation in this part is implemented as intra-eNB co- operation, a more commercially potential scheme in 3GPP. The results show OGI-aware CoMP can improve the MU-CoMP average gain from 25% to 90%.