Multi-User Networks with Outdated Channel State Information
In a communications system consisting of multiple transmitters and multiple receivers, channel state information (CSI) is necessary at the receivers for decoding of the transmitted data, and at the transmitters for managing the interference in the network. At the receivers the CSI is obtained by transmission of pilot symbols and the transmitters obtain the CSI through feedback links from the receivers. In time-varying channel, due to delays the CSI obtained by the transmitters (CSIT) can be different to the actual channel state. In the extreme case the available CSIT and the true channel state are not correlated, which means traditional interference management techniques are not applicable.
For the aforementioned delayed CSIT setting, a new interference management technique achieving good number of degrees of freedom (DoF) was recently proposed. The proposed approach is a multi-phase transmission, where in each consecutive phase the delayed CSIT from the previous phases is used to reduce the amount of the produced interference. The reduction is possible if the terms transmitted in the next phase depend on the interference terms overheard at the unintended receivers in the previous phases, which can be reconstructed at the transmitters using the delayed CSIT.
The purpose of the project is to study different types of networks under delayed CSIT setting. New transmission schemes are to be proposed, which will provide greater number of achievable DoF and/or greater sum-rate of the communication system.
Relay-Aided Communications in Large Interference Limited Networks
In multiuser wireless communication networks, three important basic network topologies can be distinguished, namely ad-hoc networks where pairs of nodes exchange information with each other, cellular networks where base stations serve multiple mobile stations, and group communication networks where more than two nodes form a group and exchange messages. For all the three network topologies mentioned above, interferences are the major performance limiting factor. Recently, interference alignment has been proposed to eliminate the interferences in such networks, especially if the interference signals and the useful signals are of comparable strengths. Besides the conventional interference alignment using time extensions or a large number of antennas at the nodes, both involving disadvantages, amplify-and-forward relays can be employed to help the process of interference alignment. By utilizing relays, the effective end-to-end channels between the communicating nodes can be manipulated in such a way, that the interferences are aligned at the receiving nodes.
In this project, we will focus on investigating relay aided interference alignment for the above three topologies in large networks, i.e., in networks containing many nodes and relays deployed over a large area. In large networks, the distances between nodes and relays differ a lot, leading to some weak channels which can be ignored. Based on this, we will consider that these large networks are partially connected. More specifically, we will consider large networks being partitioned into multiple subnetworks, which are mutually connected by a limited number of links.
Furthermore, assuming half-duplex relays, both one-way and two-way relaying protocols will be applied. One-way relaying enables using the direct links between the communicating nodes, while two-way relaying does not. On the other hand, two-way relaying needs only two transmission phases for bidirectional communication while one-way relaying needs four transmission phases. Overall, the doubled number of transmission phases required by one-way relaying is compensated by the doubled number of dimensions of the receive signal space at every destination node compared with two-way relaying. The capability to utilize or not to utilize the direct link introduces different challenges for interference alignment. Note, that we do not aim at theoretical capacity bounds in this project, but at viable solutions for the given relaying protocols.
For large networks with all the three network topologies, which may be partially connected and for both, one-way and two-way relaying protocols, we will derive the properness or feasibility conditions for interference alignment, develop interference alignment algorithms, investigate the possibilities of doing interference alignment with partial channel state information only and consider scalability issues for interference alignment in large networks.