Mattias Lampe

Design and Optimization of Wireless Sensor Network with Mobile Gateway

Conventional wireless sensor networks (WSN) convey information from a set of sensors to the desired destination by mutual collaboration in a multi-hop fashion. This paper presents an IEEE 802.15.4 standard based low power WSN with mobile gateway (MG) which for some applications has advantages over conventional WSN. The MG centric data communication network minimizes the conventional WSNs premature network partition problem i.e. separation of sink due to exhaustion of certain nodes at the stationary sinks one hop distance. Moreover a tree reconfiguration algorithm and two optimization schemes, multiple associations and hybrid tree mesh topology, are proposed in this paper. The presented approach is analyzed and validated in network simulator (NS-2).

A hybrid genetic algorithm to optimize device allocation in industrial Ethernet networks with real-time constraints

With the advance of automation technology, the scale of industrial communication networks at field level is growing. Guaranteeing real-time performance of these networks is therefore becoming an increasingly difficult task. This paper addresses the optimization of device allocation in industrial Ethernet networks with real-time constraints (DAIEN-RC). Considering the inherent diversity of real-time requirements of typical industrial applications, a novel optimization criterion based on relative delay is proposed. A hybrid genetic algorithm incorporating a reduced variable neighborhood search (GA-rVNS) is developed for DAIEN-RC. Experimental results show that the proposed novel scheme achieves a superior performance compared to existing schemes, especially for large scale industrial networks.

Multi-Objective Topology Design of Industrial Ethernet Networks

Abstract The objective of this paper is to investigate methods to guarantee the network performance of industrial Ethernet networks by an optimized topology design. The proposed approach differs from former works on network design by taking into account the specific requirements of industrial applications, the Quality of Service (QoS) related features of industrial Ethernet protocols, and the increasing network complexity in the industrial field. A multi-objective variable neighbourhood search algorithm (MOVNS) is presented offering a co-design method of physical topology and logical topology to satisfy the requirements of real-time capability, high reliability, and fast recovery time. The result analysis on simulation test cases shows the proposed algorithm can greatly improve the performance of industrial Ethernet networks at a moderate computational complexity.

Topology design of industrial ethernet networks using a multi-objective genetic algorithm

This paper addresses the problem of network topology design from an industry perspective. The topology design of industrial communication networks is formulated as a multi-objective optimization problem, considering the specific requirements of industrial applications, e.g. the diversified real-time requirements, fast recovery time, high reliability, and low cost. Different from previous publications, the problem is discussed with respect to distributed multi-ring network structures. A weighted Dandelion encoding is proposed to represent multi-ring topologies and incorporated into a multi-objective genetic algorithm to find close-to-optimal solutions. Experimental results for different application examples show the effectiveness of the proposed algorithm.

Topological design of industrial Ethernet networks with a fast heuristic

The recent rapid development of industrial Ethernet requires addressing the topic of network design from a new perspective. The design of industrial Ethernet networks must typically consider real-time constraints and robustness against network failures. In this paper, the topology design of industrial Ethernet networks problem is firstly modelled considering the requirements of industrial applications. Considering the increasing scale of industrial Ethernet networks, computational complexity of network optimization schemes becomes a more and more important issue. A fast algorithm based on a heuristic extension of a genetic algorithm is presented and shown to provide an effective solution and lower computation complexity as compared to a pure genetic algorithm.