Understanding of two-user interference channel and its application to heterogeneous networks

Abstract

In this dissertation, we newly study interference channels under three different heterogeneities -- transmission restriction via the priority in an underlay interference channel (IC), asymmetric energy consumption from energy harvesting and cooperation, and cognitive transmission with delayed channel knowledge feedback. Because of the asymmetric resources, capability, and priority to the transmission, quite different rate regions to the conventional two-user IC are investigated in each channel and efficient interference management strategies for the maximum achievable rate and corresponding maximum degrees-of-freedom (DoF) are proposed. In the underlay IC and IC with energy harvesting transmitters, Gaussian interference channel is commonly considered and the modified Han-Kobayashi (HK) scheme, which known as a near-optimal transmission strategy for the conventional two-user Gaussian IC, is studied. To find the maximum achievable rate, we derive the optimal rate split methods between common and private messages and present a closed-form expression of the corresponding power split. It was reported in literature that an asymmetric rate split between two users can maximize the sum rate in the two-user symmetric IC if no time sharing is assumed. However, we proved that the result in literature is incorrect and showed that the symmetric power split always attains the maximum sum rate. Consequently, with given optimal split powers, we find the maximum achievable rate region or corresponding DoF region in our asymmetric interference channel models.

As for the first heterogeneity to the channel, our underlay two-user IC assumes that two transmitter-receiver pairs are underlaid with one primary transmitter-receiver pair. When underlay transmitters must restrict their transmission to guarantee target DoF of a primary receiver, the achievable DoF of the underlay receivers are investigated. Either transmit power or transmit time is regulated to secure the target DoF of a primary receiver and their effects on the achievable DoF is analyzed for the simplified HK scheme. Our analytical results show that the achievable DoF of the underlay two-user IC is substantially different from that of a conventional two-user IC due to the imposed restriction on the underlay users. It is also shown that joint decoding capability at the primary receiver using common codebook relaxes the restriction imposed on the underlay users and hence improves the achievable DoF of the underlay receivers.

Secondly, we study the two-user Gaussian IC with energy harvesting and sharing. In this channel, the transmitters have different energies from random energy harvesting and energy sharing to each other. We characterize the achievable average rate region of the IC with time-varying available power when the simplified HK scheme is employed. We prove that each corner point on the average rate region is on a sum rate bound with appropriate power allocation. Based on the proof, we find the optimal strategy of energy cooperation and power allocation between the two transmitters to achieve the boundary points on the average rate region. It is shown that the energy cooperation can enlarge the average rate region compared to that of the conventional interference channel. We also show that the energy cooperation yields almost the same average rate region regardless of the interference channel condition because it enables to effectively change the given interference channel condition into a more favorable one by flexibly controlling the transmit powers.

At third, we analyze the multiple-input multiple-output (MIMO) cognitive interference channel(CIC) where one of transmitters act as a cognitive transmitter under delayed channel knowledge feedback. Specially, the receivers have current channel state information (CSI), but only delayed CSI is available at the transmitters. Because of independency of Rayleigh fading channel matrices across time, delayed CSI at the transmitters (CSIT) is outdated. With arbitrary number of transmit and receive antennas, the set of outer bounds to the DoF region are fully characterized and the contribution of the delayed channel feedback to the DoF region is studied based on that of the two-user MIMO CIC with and without perfect CSIT assumption. It is shown that the outer bound to the DoF region is achievable with the delayed CSIT while it is still an open problem without channel feedback if $M_{t,b}<M_{r,b}\leq M_{t,a}<M_{t,a}+M_{t,b}\leq M_{r,a}$ and $M_{t,b}<M_{r,b}$ where $M_{t,i}$ and $M_{r,j}$ are the number of transmit and receiver antennas at the transmitter $i$ and the receiver $j$, $i,j\in\{a,b\}$. Also, we investigate the effect of the cognitive transmission to the DoF region according to class of the cognitive networks. The cognitive transmission increases DoF region up to that of the broadcast channel (BC) for several cases under delayed CSIT. The proposed cognitive transmission scheme fully covers the outer bound to the DoF region in high signal-to-noise ratio (SNR) regime.