Capacity of interference networks : achievable regions and outer bounds

• Main Author: Sridharan, Sriram
• Format: Others
• Published: 2014
• Subjects: Interference networks; Capacity; Cognitive radios; Lattice coding
• Online Access: http://hdl.handle.net/2152/26930

In an interference network, multiple transmitters communicate with multiple receivers using the same communication channel. The capacity region of an interference network is defined as the set of data rates that can be simultaneously achieved by the users of the network. One of the most important example of an interference network is the wireless network, where the communication channel is the wireless channel. Wireless interference networks are known to be interference limited rather than noise limited since the interference power level at the receivers (caused by other user's transmissions) is much higher than the noise power level. Most wireless communication systems deployed today employ transmission strategies where the interfering signals are treated in the same manner as thermal noise. Such strategies are known to be suboptimal (in terms of achieving higher data rates), because the interfering signals generated by other transmitters have a structure to them that is very different from that of random thermal noise. Hence, there is a need to design transmission strategies that exploit this structure of the interfering signals to achieve higher data rates. However, determining optimal strategies for mitigating interference has been a long standing open problem. In fact, even for the simplest interference network with just two users, the capacity region is unknown. In this dissertation, we will investigate the capacity region of several models of interference channels. We will derive limits on achievable data rates and design effective transmission strategies that come close to achieving the limits. We will investigate two kinds of networks - "small" (usually characterized by two transmitters and two receivers) and "large" where the number of users is large. === text