Interference Management In Wireless Networks:an Information-Theoretic Perspective

Interference constitutes a major issue that limits the performance of wireless networks, especially in the scenarios of high node density and high data-rate services, which will be exactly the case in future wireless networks. Therefore, efficient interference management will play a crucial role for the performance of future wireless networks. Motivated by this, in this dissertation we are devoted to address the interference management problem in several important wireless networks, including the cellular network, cognitive network and bursty interference channel, etc. from the perspective of information theory, i.e., we are interested in characterizing the performance limit of those above-mentioned wireless networks. Especially, we’ll take the DoF (Degree of Freedom) as a major performance measure in our study, for the reason that it provides a first-order characterization of the capacity and might provide extra insight for complex channels and networks.Regarding the cellular network, we first present a full characterization of the DoF per cell for the two-cell network, which is equal to MK/M+K.This result can be seen as a mani-festation of both the multiplexing gain of multiple antennas and the multi-user gain that is typical of cellular networks. Moreover, these two gains are both based on the highly-efficient subspace-based interference alignment scheme.Next, we generalized the above result to the case of arbitrary number of cells, under which we have nearly the same DoF result as the two-cell case, namely, the DoF per cell is still MK/M+K, irrespective of the number of the cells. In addition, we can see from the result that when the number of users per cell K is large, the above DoF value will approach to M, which is exactly the maximal DoF obtained by a single cell. In other words, our scheme could achieve nearly interference-free performance (in terms of DoF) when K is large. This result is even better than the famous "half a cake per user" result by Cadambe and Jafar for the K-user interference channel, actually, in our considered scenario, we could nearly achieve the goal of "a full cake per cell".Furthermore, we consider the case of cellular networks with users equipped with multiple antennas. For this scenario, we present a DoF lower-bound and an upper-bound, which are of small gap. Moreover, when the condition G>N/gcd(M,N)(where G denotes the number of cells and gcd(M, N) denotes the greatest common divisor of M and N) is satisfied, it can be proved that the DoF lower-bound coincides the DoF upper-bound, hence providing a full DoF characterization in those scenarios.As for the causally cognitive MISO broadcast channel, we provide an achievable scheme whose essence is a combination of interference pre-cancellation that is based on the Dirty Paper Coding and interference zero-forcing. Our scheme is proved to be capable of providing considerable rate and DoF gain for the secondary system. As a by-product, we provide the capacity region of the MIMO broadcast channel with non-causal additive states known at the transmitter, which turns out to coincide the capacity region of the same channel without states at all.Our last research topic is on the busty MIMO interference channel, which aims at exploiting the statistical gain of the channel’s burstiness. For this kind of channels, we provide an exact DoF region (which includes the DoF of basic message and DoF of opportunistic message) and then apply it to the random-access interference channel to determine the best DoF operating point and the corresponding transmission schems.Overall, the above research results can be summarized as the following:First, highly-efficient interference management is crucial for the performance of wireless networks; Second, there is a very interesting interplay between multiple antennas and the interference management, that is:On one hand, the maintenance of multiplexing gain that multiple antennas could provide highly depends on the interference management scheme, if the latter performs bad, the multiplexing gain might be totally lost. On the other hand, multiple antennas are very beneficial for efficient interference management, because they could be employed to perform interference zero-forcing and interference alignment as well as help performing interference pre-cancellation.