<p>The Shannon capacity of wireless networks has a fundamental importance for network information theory. This area has recently seen remarkable progress on a variety of problems including the capacity of interference networks, X networks, cellular networks, cooperative communication networks and cognitive radio networks. While each communication scenario has its own characteristics, a common reason of these recent developments is the new idea of interference alignment. The idea of interference alignment is to consolidate the interference into smaller dimensions of signal space at each receiver and use the remaining dimensions to transmit the desired signals without any interference. However, perfect alignment of interference requires certain assumptions, such as perfect channel state information at transmitter and receiver, perfect synchronization and feedback. Today’s wireless communication systems, on the other and, do not encounter such ideal conditions. In this thesis, we cover a breadth of topics of interference alignment and cancellation schemes in wireless communication systems such as multihop relay networks, multicell networks as well as cooperation and optimisation in such systems. Our main contributions in this thesis can be summarised as follows: • We derive analytical expressions for an interference alignment scheme in a multihop relay network with imperfect channel state information, and investigate the impact of interference on such systems where interference could accumulate due to the misalignment at each hop. • We also address the dimensionality problem in larger wireless communication systems such as multi-cellular systems. We propose precoding schemes based on maximising signal power over interference and noise. We show that these precoding vectors would dramatically improve the rates for multi-user cellular networks in both uplink and downlink, without requiring an excessive number of dimensions. Furthermore, we investigate how to improve the receivers which can mitigate interference more efficiently. • We also propose partial cooperation in an interference alignment and cancellation scheme. This enables us to assess the merits of varying mixture of cooperative and non-cooperative users and the gains achievable while reducing the overhead of channel estimation. In addition to this, we analytically derive expressions for the additional interference caused by imperfect channel estimation in such cooperative systems. We also show the impact of imperfect channel estimation on cooperation gains. • Furthermore, we propose jointly optimisation of interference alignment and cancellation for multi-user multi-cellular networks in both uplink and downlink. We find the optimum set of transceivers which minimise the mean square error at each base station. We demonstrate that optimised transceivers can outperform existing interference alignment and cancellation schemes. • Finally, we consider power adaptation and user selection schemes. The simulation results indicate that user selection and power adaptation techniques based on estimated rates can improve the overall system performance significantly.</p>
Bi-LSTM based deep learning method for 5G signal detection and channel estimation
