Restore

Network Classification

On the previous pages, a technical definition of WMNs and a description of its consequent characteristics there have been given. On the following page a classification is presentes that takes into account the functional differences. The main idea here is to bring into focus the aspect of service provision and management of WMNs. In the following, three types of management are derived from the operation and control of deployment, maintenance, and administration processes found in existing WMNs.

Fully managed WMN

In such networks, all nodes providing any service are managed by a central administration. As a consequence, there are no routing mesh clients, since they would indeed offer services, namely routing. Therefore, the most apparent difference between fully managed WMNs and the general WMN depicted in the previous Figure is the lack of these routing mesh clients. Thus, a clear separation is made between the infrastructure itself, formed by backbone mesh routers and gateways only, and its users, the non-routing mesh clients. As a rule, a single company deploys, maintains, and controls the infrastructure of the WMN.

One of the most popular concept of fully managed WMNs has been advertised by Cisco Systems. Their pro jects cover entire U.S. cities and enable their customers to use the WMN as an Internet service provider (ISP) [infoweek]. The access is fully transparent to the clients, since they do not participate in the routing process, that is, clients just exchange packets with their selected AP. In other words, they are non-routing mesh clients. The remaining nodes are mesh routers and gateways that belong to, or form, the backbone.

Semi-managed WMN

In contrast to fully managed WMNs, in a semi-managed WMN just a fraction of the nodes providing any service is controlled by a single organization. This means that there is a great number of such nodes not being under control of this organization. These nodes may not only be static but also mobile and might join or leave the network at any time. Nonetheless, they are an integral part of the network, in contrast to clients in a fully managed WMN. That is, they perform tasks such as the participation in the routing process, the contribution to the autoconfiguration support, and the provision of service discovery.

A common example of a semi-managed WMN is under construction in the city of Berlin. At the moment, up to 400 routing nodes of the “Freifunker” community [freifunker] cover parts of the city. These include both nodes that are administrated by the community organization as such and nodes that are managed individually. Some of the community nodes have got antennas mounted on high places in order to bridge long distances and, thus, to minimize the effective hop count. All nodes of the WMN participate in routing, including mobile and dynamic user nodes. Hence, the network topology may change frequently in semi-managed WMNs while it stays rather static in fully managed WMNs.

Unmanaged WMN

In an unmanaged WMN, there is nothing like a central administration that manages stable and service providing resources. Instead, such networks are formed by nodes that are under control of individual enterprises or persons. Similar to MANETs, there is no dedicated, pre-existing infrastructure required. An unmanaged WMN adopts the available infrastructure and, therefore, might persist only for a transitory period of time. However, nodes of a MANET are supposed to move more frequently achieving a higher mobility. In contrast, nodes in an unmanaged WMN move less often and, hence, they are rather static. An unmanaged WMN may emerge, e.g., in a conference room or on a university campus. Moreover, some nodes may act as gateways connecting the WMN to external networks like the Internet, e.g., via GSM, UMTS, or WLAN.