Coding strategies for relay based networks

For future wireless communication networks one of the major concerns for service providers is to provide seamless connectivity to the end-users with quality of service (QoS) as high as possible. However, to achieve the determined QoSs for all users in the network is a challenging issue due to the time-varying characteristics of communication channels, caused by multi-path fading, path-loss and shadowing. Recently, base station (BS) cooperation and relay station (RS) deployment have been proposed as promising technologies to notably improve the performance of next-generation wireless systems in terms of fairness, coverage, energy/cost and spectral efficiency.

In this thesis, we gravitate our attention towards the use of RSs in different wireless communication systems such as cellular, Ad-hoc and satellite networks with reliability and achievable rates being our main figures of merit. In particular, we project the insights gained from information theoretic analysis of various relaying strategies into the real world settings, and assess the effectiveness and potentials of relaying in various wireless applications.

In the first part of the thesis, we focus on parallel relay networks (PRNs) and investigate whether it is possible to have good performance by using simple and cheap RSs for PRNs with limited backhaul connections to the destination. In particular, we propose a simple and practical quantization technique at the RSs which relies on symbol-by-symbol uniform scalar quantization (uSQ). For the same network model, we also characterize the random coding error exponents (EEs) corresponding to different relaying strategies used by the RSs. We show that in certain regimes the EEs achieved by simple relaying strategies are better than the EEs of more complex counterparts.

Inspired by the PRN setup, in the second part of the thesis, we consider cellular networks assisted by fixed RSs for both uplink (UL) and downlink (DL) communications. In particular, we analyze achievable sum-of-rates for UL communications assuming that mobile station (MS) and RS signals are emitted on orthogonal frequency bands, and thoroughly assess the influences of RS deployment on the achievable sum-of-rates considering several parameters used by the system such as the number of RSs deployed in each cell, their locations and powers they use. For the relay-assisted cellular DL communications, we propose a two-step distributed scheduling algorithm where a given MS can be served by either the BS or one of the deployed RSs, in an opportunistic way. Such a distributed approach allows a reduced feedback signaling with respect to the centralized case, especially when a simple scalar feedback is not sufficient for estimating the channel quality.

In the last part of the thesis, we propose various coding strategies for two-way relay channels (TWRCs). Specifically, we show that lattice-based partial decoding strategies achieve near optimal performance. However, since this coding scheme highly relies on the channel impairments, it might not find applications in real scenarios. To this end, we demonstrate that the use of binning-type relaying strategies might be a more practical approach since phase coherence of the signals at the RS is not required. Finally, we extent the single-pair TWRC model to a multi-pair case where in each pair the single-antenna MSs seek to communicate via a common multiple-antenna RS. In the multi-pair TWRC, the main bottleneck on system performance is the interference seen by each MS due to the other communicating MS pairs. To tackle this problem, we propose some spatial domain solutions by exploiting multiple antennas at the RS.