Amila Pradeep Kumara Tharaperiya Gamage
Resource Management for Heterogeneous Wireless Networks
Demand for high volumes of mobile data traffic with better quality-of-service (QoS) support and seamless network coverage is ever increasing, due to growth of the number of smart mobile devices and the applications that run on these devices. Seamless network coverage can be achieved by interworking of heterogeneous wireless networks to merge coverage of individual networks. To support high volumes of data traffic with better QoS, more frequency spectrum can be added or spectrum efficiency enhancing techniques can be applied to the networks. However, available frequency spectrum that can be used for outdoor robust communications is limited, as most of the underutilized spectrum is in the high frequency range. Therefore, spectrum efficiency enhancing techniques, such as, interworking, device-to-device (D2D) communication, frequency reuse, and densely deploying small cells, should be used in the networks. Since a single technique cannot enhance the spectrum efficiency throughout a network, few of these techniques can be combined to achieve better performance. However, one of the challenging issues that need to be solved before these techniques can be applied in practical networks is the efficient resource management. Efficiently managing resources, such as frequency and transmit power, is crucial as it has a direct impact on the spectrum efficiency. Therefore, this thesis focuses on studying and developing efficient resource management schemes for heterogeneous wireless networks.
First, uplink resource allocation for cellular network and wireless local area network (WLAN) interworking to provide multi-homing voice and data services is investigated. The main technical challenge, which makes the resource allocation for this system complicated, is that resource allocation decisions need to be made capturing multiple physical layer (PHY) and medium access control layer (MAC) technologies of the two networks. This is essential to ensure that the decisions are feasible and can be executed at the lower layers. Thus, the resource allocation problem is formulated based on PHY and MAC technologies of the two networks. The optimal resource allocation problem is a multiple time-scale Markov decision process (MMDP) as the two networks operate at different time-scales, and due to voice and data service requirements. A resource allocation scheme consisting of decision policies for the upper and the lower levels of the MMDP is derived. To reduce the time complexity, a heuristic resource allocation algorithm is also proposed.
Second, resource allocation for D2D communication underlaying cellular network and WLAN interworking is investigated. Enabling D2D communication within the interworking system further enhances the spectrum efficiency, especially at areas where only one network is available. In addition to the technical challenges encountered in the first interworking system, interference management and selection of users' communication modes for multiple networks to maximize hop and reuse gains complicate resource allocation for this system. To address these challenges, a semi-distributed resource allocation scheme that performs mode selection, allocation of WLAN resources, and allocation of cellular network resources in three different time-scales is proposed.
Third, resource allocation for interworking macrocell and hyper-dense small cell networks is studied. Such system is particularly useful for interference prone and high capacity demanding areas, such as busy streets and city centers, as it uses license frequency bands and provides a high spectrum efficiency through frequency reuse and bringing network closer to the users. The key challenge for allocating resources for this system is high complexity of the resource allocation scheme due to requirement to jointly allocate resources for a large number of small cells to manage co-channel interference (CCI) in the system. Further, the resource allocation scheme should minimize the computational burden for low-cost small cell base stations (BSs), be able to adapt to time-varying network load conditions, and reduce signaling overhead in the small cell backhauls with limited capacity. To this end, a resource allocation scheme which operates on two time-scales and utilizes cloud computing to determine resource allocation decisions is proposed. Resource allocation decisions are made at the cloud in a slow time-scale, and are further optimized at the BSs in a fast time-scale in order to adapt the decisions to fast varying wireless channel conditions. Achievable throughput and QoS improvements using the proposed resource allocation schemes for all three systems are demonstrated via simulation results.
In summary, designing of the proposed resource allocation schemes provides valuable insights on how to efficiently allocate resources considering PHY and MAC technologies of the heterogeneous wireless networks, and how to utilize cloud computing to assist executing a complex resource allocation scheme. Furthermore, it also demonstrates how to operate a resource allocation scheme over multiple time-scales. This is particularly important if the scheme is complex and requires a long time to execute, yet the resource allocation decisions are needed to be made within a short interval.