Self-Organizing Network Introduction

Ad-hoc networks are a hybrid of mobile communication and computer networks, where information exchange employs packet switching mechanisms from computer networks. User terminals are portable and movable, with each terminal in the ad-hoc network serving both router and host functions. As a host, the terminal must run various user-facing applications such as editors and browsers; as a router, it must run the corresponding routing protocols, forwarding data packets and maintaining routes based on routing strategies and tables. Therefore, nodes must implement appropriate routing protocols. The goal of ad-hoc network routing protocols is to be fast, accurate, and efficient, requiring the quick discovery of accurate and available routing information within the shortest possible time, while adapting to rapid changes in network topology. Additionally, they aim to minimize extra latency and control information maintenance for routing, reducing the overhead of routing protocols to accommodate the limitations of mobile terminals in terms of computational power, storage space, and power.

The design of ad-hoc network routing protocols primarily involves three approaches: (1) Modify existing conventional routing protocols to meet the needs of ad-hoc networks, such as the DSDV (Destination Sequenced Distance Vector) protocol, which is derived from the common RIP protocol; (2) Adopt on-demand routing principles, avoiding periodic broadcast of routing information to maintain routing tables. Requests are only made to establish routes when needed, effectively reducing resource consumption. Examples include Dynamic Source Routing (DSR) and AODV (Ad-hoc On-demand Distance Vector); (3) QoS-based routing, where nodes select a route that is most likely to meet user QoS requirements based on collected network resource information (instead of the usual hop count), such as LS-QoS (link State-QoS Protocol

Table-driven routing protocols are suitable for conventional wired networks, but for wireless ad-hoc networks, due to numerous inherent limitations, periodic broadcast control information packets can consume a significant amount of network bandwidth, while maintaining routing tables can consume substantial resources of mobile terminals. Rapid changes in the topology structure can render many routing information outdated quickly, leading to resource wastage. Even with modifications tailored for wireless ad-hoc networks, several prevalent on-demand routing protocols remain largely in use. DSR employs a source routing mechanism, requiring complete path information in each packet header, significantly increasing the overhead of the routing protocol. Moreover, during link failures that necessitate route reconstruction, the broken link information must be sent back to the source node, causing a re-initiation of the route discovery process by the source node, resulting in considerable delays. AODV resolves this issue using a hop-by-hop forwarding mechanism, but it requires periodic Hello messages to maintain node connection states, adding to the overhead, and in the event of a link failure, it rebuilds routes in a similar manner to DSR. The TORA protocol, aside from its high overhead, also requires specialized hardware support, such as GPS devices for time synchronization across all nodes and the need for two separate wireless channels for data and control, greatly limiting its application scope.

The characteristics of self-organizing networks include:

Equivalence: All nodes have equal roles within the network, allowing them to freely join or leave the network without notifying other nodes, without affecting the overall stability of the network.

Dynamic Topology Structure: Due to nodes' ability to freely move and start/stop, the topology of self-organized networks is typically dynamic, supporting various networking modes such as point-to-point, star, and mesh configurations.

No central control: In ad-hoc networks, there is no central controller; the behavior of all nodes is coordinated through layered protocols and distributed algorithms.

Resilience: Due to the independence and equality among nodes, a failure of a node in an ad-hoc network does not affect the overall operation of the network.