Algorithm Design for Scheduling and Medium Access Control in Heterogeneous Mobile Networks

Abstract

The rapid growth of wireless mobile devices has led to saturation and congestion of wireless channels – a well-known fact. In the recent years, this issue is further exacerbated by the everincreasing demand for traffic intensed multimedia content applications, which include but are not limited to social media, news and video streaming applications. Therefore the development of highly efficient content distribution technologies is of utmost importance, specifically to cope with the scarcity and the high cost of wireless resources. To this aim, this thesis investigates the challenges and the considerations required to design efficient techniques to improve the performance of wireless networks. Since wireless signals are prone to fluctuations and mobile users are, with high likelihood, have difference channel qualities, we particularly focus on the scenarios with heterogeneous user distribution. Further, this dissertation considers two main techniques to cope with mobile users demand and the limitation of wireless resources. Firstly, we propose an opportunistic multicast scheduling to efficiently distribute or disseminate data to all users with low delay. Secondly, we exploit the Millimeter-Wave (mm-Wave) frequency band that has a high potential of meeting the high bandwidth demand. In particular, we propose a channel access mechanism and a scheduling algorithm that take into account the limitation of the high frequency band (i.e., high path loss). Multicast scheduling has emerged as one of the most promising techniques for multicast applications when multiple users require the same content from the base station. Unlike a unicast scheduler which sequentially serves the individual users, a multicast scheduler efficiently utilizes the wireless resources by simultaneously transmitting to multiple users. Precisely, it multiplies the gain in terms of the system throughput compared to unicast transmissions. In spite of the fact that multicast schedulers are more efficient than unicast schedulers, scheduling for multicast transmission is a challenging task. In particular, base station can only chose one rate to transmit to all users. While determining the rate for users with a similar instantaneous channel quality is straight forward, it is non-trivial when users have different instantaneous channel qualities, i.e., when the channel is heterogeneous. In such a scenario, on one hand, transmitting at a low rate results in low throughput. On the other hand, transmitting at a high rate causes some users to fail to receive the transmitted packet while others successfully receive it but with a rate lower than their maximum rate. The most common and simplest multicasting technique, i.e., broadcasting, transmits to all receivers using the maximum rate that is supported by the worst receiver. In recent years, opportunistic schedulers have been considered for multicasting. Opportunistic multicast schedulers maximize instantaneous throughput and transmit at a higher rate to serve only a subset of the multicast users. While broadcasting suffers from high delay for all users due to low transmission rate, the latter causes a long delay for the users with worse channel quality as they always favor users with better channel quality. To address these problems, we designed an opportunistic multicast scheduling mechanism that aims to achieve high throughput as well as low delay. Precisely, we are solving the finite horizon problem for multicasting. Our goal is that all multicast users receive the same amount of data within the shortest amount of time. Although our proposed opportunistic multicast scheduling mechanism improves the system throughput and reduces delay, a common problem in multicast scheduling is that its throughput performance is limited by the worst user in the system. To overcome this problem, transmit beamforming can be used to adjust antenna gains to the different receivers. This allows improving the SNR of the receiver with the worst channel SNR at the expense of worsening the SNR of the better channel receivers. In the first part of this thesis, two different versions of the finite horizon problem are considered: (i) opportunistic multicast scheduling and (ii) opportunistic multicast beamforming. In recent years, many researchers venture into the potential of communication over mm-Wave band as it potentially solves the existing network capacity problem. Since beamforming is capable to concentrate the transmit energy in the direction of interest, this technique is particularly beneficial to improve signal quality of the highly attenuated mm-Wave signal. Although directional beamforming in mm-Wave offers multi-gigabit-per-second data rates, directional communication severely deteriorates the channel sensing capability of a user. For instance, when a user is not within the transmission coverage or range of the communicating users, it is unable to identify the state of the channel (i.e., busy or free). As a result, this leads to a problem commonly known as the deafness problem. This calls for rethinking of the legacy medium access control and scheduling mechanisms for mm-Wave communication. Further, without omni-directional transmission, disseminating or broadcasting global information also becomes complex. To cope with these issues, we propose two techniques in the second part of this thesis. First, leveraging that recent mobile devices have multiple wireless interface, we present a dual-band solution. This solution exploits the omni-directional capable lower frequency bands (i.e., 2.4 and 5 GHz) to transmit control messages and the mm-Wave band for high speed data transmission. Second, we develop a decentralized scheduling technique which copes with the deafness problem in mm-Wave through a learning mechanism. In a nutshell, this thesis explores solutions which (i) improve the utilization of the network resources through multicasting and (ii) meet the mobile user demand with the abundant channel resources available at high frequency bands.