Design of Survivable Wireless Backhaul Networks Chalermpol Charnsripinyo David Tipper Department of Information Science and Telecommunications University of Pittsburgh Pittsburgh, PA 15260 USA Email: charnsri, [email protected]
Keywords: Wireless access networks, Mobile cellular networks, Network design, Network survivability
1. Introduction Wireless access networks have become crucial to provide mobile users with untethered communication services regardless of location, mobility pattern, or type of terminal used. Typical wireless access networks include analog and digital cellular phone networks, Personal Communication Services (PCS) networks, wireless local area networks, and wide-area mobile data service networks (e.g., General Packet Radio Service in GSM systems, etc.). Among these, mobile cellular/PCS networks currently represent the rapid growing sector with current emphasis on mobile data services. As societal dependence on mobile services increases, especially for emergency services, network failures that inhibit communications or results in loss of critical data cannot be tolerated. Many human-made and natural causes (e.g., fires, tornados, floods, earthquakes, construction or terrorist activities) can result in network failures and disrupt communication services. Therefore, it is essential to take into account such failure scenarios and their potential effects when designing mobile cellular networks. In general, survivable networks can be realized by designing of network infrastructures that are robust to link or equipment failures, and implementing control systems that are inherently fault-tolerant and selfhealing. However, cost is usually a key factor in the network design. Allocating spare resources to provide protection against component failures means a greater cost in implementation. Due to the complexity of wireless network architecture, unique characteristics of wireless mobile networks such as user mobility, and quality of service (QoS) requirements of voice and emerging data services in mobile cellular systems, the survivable wireless network design can be costly and highly complicated. The challenge is to design a costeffective wireless network to meet user demands with acceptable reliability and quality of services.
2. Survivable Network Design This paper proposes a network design model for survivable wireless backhaul networks. The main objective is to design minimum-cost wireless networks that can support voice and data services while satisfying QoS and survivability requirements. This involves three main tasks which are designing faulttolerant network topology, dimensioning links between network components, and routing traffic demand subject to survivability requirements. The survivable wireless backhaul network design problem has a number of unique issues which differentiate it from conventional wired backbone networks design (i.e., public switched telephone networks, and data packet switched networks). A typical wireless backhaul network consists of a collection of network components or nodes (i.e., base stations, base station controllers and mobile switching centers) and links connecting between these components. Each base station (BS) or cell site generates traffic which represents demand of the BS to be carried in the backhaul network. These traffic demands carried through links between BSs and BSC (or hub) are aggregated into higher capacity links and routed to the mobile switching center (MSC). This traditional interconnect backhaul network has a tree-like (root-branch-leaf) topology with the MSC at the root. The tree network topology could be the least-cost network design due to the minimum connectivity and the advantage of economies of scale in the cost of higher capacity but it does not provide protection against a single link failure. There are several strategies for designing survivable network topologies. In the tree-like topology, additional connectivity and automatic protection switches are required in the network topology to achieve
an automatic recovery from a single failure. Other network topologies like self-healing ring and mesh topologies are capable of rerouting traffic around the point of failure through other existing facilities and switches. A big question is which type of network topology is the most cost-effective in a survivable network design. Note that in the wireless backhaul network design, the network is organized into a facility hierarchy. That is, each BS is controlled by a BSC and all BSCs are connected directly or indirectly with the MSC. This hierarchy tends to suggest that traffic be concentrated into high capacity routes to central locations. In mobile data services (e.g., General Packet Radio Service, GPRS) and third-generation cellular networks, additional network components (e.g., serving GPRS support node, and gateway GPRS support node) need to be considered in the network design as well as different traffic types and QoSs should be considered. Another important issue that needs to be considered in a survivable wireless backhaul network design is the user mobility. Unlike wired networks, user mobility in wireless mobile networks can significantly worsen network performance after failures, as disconnected users move among adjacent cells and attempt to reconnect to the network . Therefore, the capacity requirement for network dimensioning is different in wireless backhaul networks when compared to wired networks due to the movement of users. Other factors in wireless networks like the location and shape of failed area should be taken into account for a survivable wireless network design. For example, the impact of failures in cell sites serving mobile users on highway can be worse than those in residential areas. The importance of providing fault tolerance in wireless backhaul networks together with unique characteristics of wireless access networks has necessitated the development of new network design models. Our approach is to formulate the problem of survivable wireless backhaul network design as a mixed integer-programming model including those aspects previously mentioned. The objective of the MIP model is to minimize the cost of fault-tolerant topology design that incorporates the placement of spare network capacity for mitigating the impacts of component failures and providing continuity of services with acceptable QoS level to mobile users during failures. A piecewise-linear cost of network capacity is considered. This allows the network to take advantage of economies of scale in the cost of network capacity resulting in more realistic and cost-effective implementation. Initially, the current secondgeneration mobile cellular networks supporting voice services are considered. The survivability requirement would be the percentage of restorable network traffic upon certain failure scenarios. The QoS requirement would be the call blocking probability. A pre-computed hop count limited path set is used in the network design model. The model will determine the network topology, dimension of links, and optimal routes such that survivability and QoS requirements are met, given that the traffic demand, location of each base station and mobile switching center are known. In addition, the model will resolve the number of base station controllers needed and their locations in the network topology. Different types of network topologies including the tree-like topology with diverse paths, the self-healing ring topology as the backbone, and the mesh topology are compared in terms of total network cost and the percentage of network redundancy. The solution of survivable wireless backhaul network design will depend on several factors including the number of BS or cell cites, traffic loads, cost models of facilities, survivability and QoS requirements. A set of sample problems is made to compare the results. The model described above focuses on voice services in the current mobile cellular networks. For the design of survivable wireless data networks, several issues including type of services, QoS requirements and characteristics of data services must be considered. Extension to the model for next generation networks supporting voice, data and emerging multimedia services will be discussed.
References  D. Tipper, T. Dahlberg, H. Shin, and C. Charnsripinyo, “Providing Fault Tolerance in Wireless Access Networks,” IEEE Communications Mag., Jan 2002, pp. 2-8.