Multi-Hub Location Heuristic for Alert Routing

Abstract

Trigger warning and other delay-sensitive applications for environmental, healthcare, and industrial or military surveillance and monitoring are usually based on networks. Surveillance nodes (like in the Internet of Things or wireless sensor networks) send the data to selected nodes (hubs) that forward the alert or alarm further to the next system level. The maximum length of the shortest paths from the nodes to their nearest hub should be optimized to minimize the maximum necessary number of hops required to route a warning. For k hubs, this requirement is expressed as k-source minimization of the maximum vertex eccentricity problem, i.e. minimization of d in a d-hop dominating set of a given cardinality k. In this contribution, several heuristic algorithms selecting the initial set H of hubs (i.e. the dominating set) are combined with our greedy and k-means-like approaches adapted from Euclidean to graph (i.e. geodesic) distance. The presented algorithms were tested on static geometric network models, standardly used for models of Wi-Fi networks. The best results were produced by the k-means-like algorithm with initialization provided by the centers of communities found by edge-betweenness community detection. When compared with, e.g., random selection of hub locations, our method can cut the worst case number of hops by half, and when compared with a classical k center approach, our improvement was more than 50%.INDEX TERMS Clustering methods, communication networks, heuristic algorithms, optimization.

Heuristics for Spreading Alarm throughout a Network

“…Recently the problem has been optimized by stochastic metaheuristics like tabu search or simulated annealing [11], paying for better results with computational time. Instead of spreading an alert through seed hubs, this problem can be interpreted as selecting hubs for sending an alert to a higher level of the network, so that an alert occurring in any node of the network is sent through the closest hub in the shortest time [12]. The alert routing can be employed e.g., in security [13].…”

Marine Observation Beacon Clustering and Recycling Technology Based on Wireless Sensor Networks

“…The Dijkstra algorithm was used to obtain the optimal path for information transfer. However, GA is computationally intensive and fuzzy rules apply simple if–then rules to distribute computing whether it is machine learning, other intelligent algorithms, or natural-based heuristic evolutionary algorithms [43], which require nodes and BSs with high computing performance and high memory capacity. In particular, when the number of networks is large, the system delay increases [26].…”