This chapter presents an architecture for context-aware Virtual Networks (VNs) that provides user-driven multi-access communication. The architecture is dedicated and appropriate for the flexibility provided by Wireless Mesh Networks. According to this architecture, VNs with different context characteristics such as Quality of Service (QoS), mobility, and security are built to support communications with different characteristics that best fit the users’ needs. The architecture is modeled both through a probabilistic and an optimization approach to provide quantitative insights into its performance. The probabilistic model quantifies the overhead on the architecture in terms of networking delays induced due to the VN management (searching, creation, and management). The optimization model provides insights into the competition of the VNs for the limited bandwidth resources. Indicative results of the models show the feasibility of the architecture, the upper bounds in terms of number of supported VNs to achieve good quality communications, and the relative placement of the flows in different, competing VNs.
Wireless Mesh Networks (WMNs) are spreading both in city and rural areas and connect users in both home and mobile environments, in company areas, campuses, and a variety of other settings. The term WMN includes a variety of networks showing a mesh topology and connecting over wireless links. In this chapter, we concentrate on larger WMNs based on a variety of wireless technologies (multi-access) covering larger areas which can make best use of virtualization. As most operational WMNs consist of fixed mesh routers and fixed and mobile mesh clients, this is the setting we refer to in our further investigations. One typical example are community provided WMNs for providing Internet access in urban areas. The topology of the WMNs’ physical infrastructure provides high flexibility and can support changing and switching routes and transport connections, which can be exploited to provide multi-access communication and support for a variety of different services and situations. Context information can be used to autonomously build logical networks on top of physical networks. By context we mean any information characterizing the situation of the user, including his/her services and preferences, such as Quality of Service (QoS) requirements, pricing preferences, mobility behaviors, and privacy and security demands. The logical networks are formed based on context information expressing user requirements, and are described by the context characteristics each of them supports. Users are grouped according to similarity of their environmental conditions and preferences. User groups can be assigned to the logical networks matching their context.
When building logical networks, network virtualization can be a very useful technique for a flexible utilization of a physical network infrastructure. For instance, wireless mesh links are prone to disturbances on the air interface, such as reflection and absorbing of the signal due to weather changes, new buildings, car traffic conditions, etc. A flexible solution which is capable to adapt and make use of different wireless link qualities changing over time is required (e.g. operating on different frequencies and power levels). Network virtualization is able to make parallel use of the same infrastructure network and even combine physical network components to a single logical component. Hence, network virtualization is a powerful tool to build multiple logical networks and, thus, to support different network organizations and purposes. In combination with network federation (cooperation of different network operators and parties), network virtualization is seen as a major step for managing larger sets of physical resources to provide better services.
Bringing WMNs, context-awareness, and network virtualization together, this chapter presents a novel architecture for WMNs that consists of a multitude of context-based Virtual Networks (VNs), each one providing personalized communications for a user or group of users, according to their requirements, which are part of the user context. This approach is able to provide user-driven multi-access communications, since a user can simultaneously access different VNs in the same infrastructure according to the services and context associated with each VN. In the presented approach, we model user context in terms of QoS, security, user mobility, cost, etc., and virtualize the WMN by introducing multiple VNs, each corresponding to a specific context. The mesh topology, together with the WMNs’ self-organizing characteristics, allows for the introduction of virtualization following different approaches. First, the physical resources of a single WMN can be shared by different VNs, and, second, resources belonging to multiple WMNs can be aggregated to a single VN. For example, multiple paths owned by different providers can be aggregated providing increased reliability. Hereby, we further consider and include multi-homing supported by different providers and different WMNs in our approach.
This approach raises several challenges that need to be studied, such as: (i) the characterization of VNs according to their context; (ii) the discovery of the best fitting VN for a specific user, according to the context of all entities involved; (iii) the match of available context VNs against the users’ context requirements; and (iv) the VN topology maintenance and resource management, given that the VNs are built on top of the physical networks.