Network Routing Information Research Articles

Application of Q‐Learning in Routing of Software‐Defined Wireless Mesh Network

AbstractA wireless mesh network uses an ad‐hoc routing module for communication between nodes by referring to the addresses assigned to the nodes according to the rules kept in the routing table. If links fail, then the necessary network routing information must be propagated via the relevant remaining reachable nodes. The weaknesses of this conventional wireless mesh network are its difficulty in terms of management and configuration adjustments due to the complex structure of the network. To cope with such difficulty, we propose the use of a machine learning technique to implement dynamic routing to deal with failures in the wireless mesh network. The aim is to improve network routing via a reinforcement‐learning framework. Specifically, the Q‐learning algorithm has been applied. The investigation results are reported for network scenarios under conditions of link failures, and the algorithm's parameter sensitivity analysis is given. In addition, the proposed algorithm is compared with the Dijkstra algorithm with either no or immediate link cost updates. The resultant end‐to‐end route latency is reported from the simulated conditions as parameterized by the real per‐hop latency measurement from the actual outdoor testbed implementation. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

Pworm: Passive and Real-Time Wormhole Detection Scheme for WSNs

Wormhole attack is one of the severe threats to wireless sensor and ad hoc networks. Most of the existing countermeasures either require specialized hardware or demand high network overheads in order to capture the specific symptoms induced by the wormholes, which in result limits their applicability. In this paper, we exploit an inevitable symptom of wormholes and present Pworm, a passive wormhole detection and localization system based upon the key observation that a large amount of network traffic will be attracted by the wormholes. The proposed scheme is passive, real-time, and efficient against both active and passive wormholes. Our approach silently observes the variations in network topology to infer the wormhole existence and solely relies on network routing information. It does not necessitate specialized hardware or poses rigorous assumptions on network features. We evaluate our scheme through extensive simulations of 100 to 800 nodes for various network scales and show that Pworm is well suited for false alarms, scalability, and time delay in terms of activation as well as detection latencies.

An improved ant colony system algorithm for solving the IP traceback problem

The difficulty in identifying the origin of an attack over the Internet is termed the IP traceback (IPTBK) problem. The probable origin of an attack is commonly investigated using some form of ant colony system (ACS) algorithms. However, such algorithms tend to converge to a local suboptimal solution, meaning that the perpetrator of the attack cannot be found. Therefore, the present study proposes a modified ACS scheme (denoted as ACS–IPTBK) that can identify the true attack path even without the entire network routing information. The ability of the ants to search all feasible attack paths was enhanced using a global heuristic mechanism in which the ant colony was partitioned into multiple subgroups, with each subgroup having its own pheromone updating rule. The performance of the ACS–IPTBK algorithm in reconstructing the attack path was investigated through a series of ns2 simulations by using network topologies generated by the Waxman model. The simulations focused specifically on the effects of the ACS model parameters and network characteristics on the performance of the ACS–IPTBK scheme in converging towards the true attack path. Finally, the robustness of the proposed scheme against spoofed IP attacks was investigated. The results showed that the proposed scheme has a slightly slower convergence speed than does the conventional ACS algorithm, but yields a more globally optimal solution for the attack path, particularly in large-scale network topologies.

A novel architecture of optical code label generation and recognition for optical packet switching

A novel architecture of optical code (OC) label generation and recognition for optical packet switching (OPS) by using super structured fiber Bragg grating (SSFBG) is proposed. The OC label is generated and recognized by a label generator and recognizer, respectively. The label generator is composed of N encoders in parallel, and it can generate 2N kinds of serial optical code labels (SOCLs) for indicating 2N network routing information. The label recognizer can decode SOCLs by N decoders in parallel and provides label information to the switching control unit so that clock information is not required during the decoding process. In the switch nodes, handling of the high-speed information payload stream and the recognition of the OC label are performed in the optical domain, while processing of the routing information remains in the electrical domain. This approach could be a promising solution for an OPS network with high capacity, good quality of service (QoS), multi-service function and high security. In this experiment, we demonstrate 40 Gbps 256 label optical packet switching that employs clockless SOCL processing.

A tree convolution algorithm for the solution of queueing networks

A new algorithm called the tree convolution algorithm, for the computation of normalization constants and performance measures of product-form queueing networks, is presented. Compared to existing algorithms, the algorithm is very efficient in the solution of networks with many service centers and many sparse routing chains. (A network is said to have sparse routing chains if the chains visit, on the average, only a small fraction of all centers in the network.) In such a network, substantial time and space savings can be achieved by exploiting the network's routing information. The time and space reductions are made possible by two features of the algorithm: (1) the sequence of array convolutions to compute a normalization constant is determined according to the traversal of a tree; (2) the convolutions are performed between arrays that are smaller than arrays used by existing algorithms. The routing information of a given network is used to configure the tree to reduce the algorithm's time and space requirements; some effective heuristics for optimization are described. An exact solution of a communication network model with 64 queues and 32 routing chains is illustrated.