With the rapid development of the mobile communication and wireless technologies, the demand for high throughput and continuous connectivity are also increasing day by day. Relay-assisted communication is the fundamental technique to be used to provide coverage extension and capacity increase in wireless networks. However, in order to achieve the optimal performance, the maximum-likelihood detector is needed while it has the exponential complexity and it is generally not practical. This thesis describes a simplified maximum-likelihood detector for the three-node relay channel which is derived from the geometric insight of the decision boundaries in the signal space. We show that this simplified maximum-likelihood detector can perform close to the actual one with linear complexity. On th...[ Read more ]

With the rapid development of the mobile communication and wireless technologies, the demand for high throughput and continuous connectivity are also increasing day by day. Relay-assisted communication is the fundamental technique to be used to provide coverage extension and capacity increase in wireless networks. However, in order to achieve the optimal performance, the maximum-likelihood detector is needed while it has the exponential complexity and it is generally not practical. This thesis describes a simplified maximum-likelihood detector for the three-node relay channel which is derived from the geometric insight of the decision boundaries in the signal space. We show that this simplified maximum-likelihood detector can perform close to the actual one with linear complexity. On the other hand, network coding can be naturally applied in multi-terminal networks which can improve the network capacity. In the literature, the network coding vector is inserted into the packet to keep track of the linear combination of the source packets the network coded packet contains but it may incur too much overhead in some application scenarios. Compression of network coding vector is a more bandwidth-efficient way to realize network coding. We propose an error-and-erasure decoding scheme to enrich the possible choices of compression schemes in the literature. The proposed scheme can also provide the flexibility to adapt the instantaneous estimation of the maximum allowable number of source packets to be combined at the intermediate node to reduce the processing complexity and increase the compression ratio.