MIMO interference networks : DoF characterization and interference analysis

MIMO 干擾網絡 : 自由度刻畫以及干擾分析

Student thesis: Doctoral Thesis

View graph of relations

Author(s)

  • Jiayi CHEN

Related Research Unit(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date15 Jul 2013

Abstract

In this thesis, we propose a new method to address the problem of characterizing the degree-of-freedom (DoF) region achievable by interference alignment for multi-user multi-antenna interference channels. This method is then applied to study DoF of multi-antenna interference broadcast channels. Besides DoF characterization, we also investigate statistical properties of aggregate interference induced in a finite ad hoc network and the impact of interference on network performances. Interference alignment has been shown to achieve the DoF capacity of the multi-user single-input-single-output (SISO) interference channel, suggesting that it is a promising technique in managing interference. Due to the good performance, it is very desirable to apply this technique to multi-user multiple-input-multiple-output (MIMO) interference channels. The concerned problem is what is the achievable DoF in a MIMO interference channel, which is still unknown in general to date. To address this problem, we propose a novel approach - joint space-decomposition. This approach provides an unifying method to derive the DoF region achieved by spatial interference alignment in an interference channel with arbitrary numbers of transmit and receive antennas installed in each user. By this approach, we obtain exact achievable DoF regions for the two-user and three-user MIMO interference channels. For the case with more than three users, we formulate it as an optimization problem of maximizing the dimension of the intersection subspace of some subspaces, based on which we develop an inner and an outer bound for the DoF region. Interference alignment can also be applied to cellular networks to mitigate intercell interference. In such networks, we derive inner and outer regions bounding the achievable DoF regions from inside and outside for the two-cell and three-cell cases, where each cell serves two users. Examples are provided to illustrate the average gap between the inner and the outer bounds to be not greater than one DoF per dimension for the two-cell case. In some networks such as ad hoc networks where interference cannot be eliminated, interference analysis have to be conducted rather than DoF characterization. We study the co-channel interference arising from uncoordinated transmissions in ad hoc networks by characterizing its statistics and then investigate its impacts on link/system performances. Different from existing works where an infinite number of nodes distributing on an infinite plane is assumed, we study a finite ad hoc network network with a finite number of users spreading over a finite area. Accordingly, the binomial point process is employed to model the dynamic positions of the nodes, instead of the ubiquitously used Poisson point process. All nodes in the network are equipped with multiple antennas. At transmitters, either spatial multiplexing or antenna selection is employed, depending on the availability of channel state information feedback. At receivers, maximal ratio combining and zero forcing receptions are adopted. In such a network, the moment generating functions of the aggregate interference received by a test receiver are derived, based on which the moments of interference are obtained in closed-form. In addition, the outage probability of a test link can be derived. Diversity analysis of the outage probability shows that the full diversity provided by the channel can be achieved by the single-stream transmission, including transmit antenna selection. The network-wide performance metric of average network throughput is also analyzed, showing that maximum throughput can be achieved by compromising between number of spatially multiplexed streams and the link outage probability.

    Research areas

  • Wireless communication systems, MIMO systems