Graham McMahon

Genetic algorithms in the design of complex distribution networks

Reviews the value of network concepts as a means of portraying complex logistics and distribution systems. Reports on research which focuses on the broader issues of model formulation and solution techniques rather than specific applications. Addresses the issues of designing networks with a tree structure, and also more general ones in which loops are allowed and redundancy enforced. The decision variables involved are related to whether or not a link should exist between two specific pairs of nodes, and then what should be the level of traffic flow on that particular link. Describes the design problem in detail and possible models that could be used to represent it. Follows with a description of genetic algorithms to solve the synthesis problem of deciding the node‐link topology, and the use of linear and non‐linear programming to solve the problem of assigning traffic flow to a network with a given typology in a least‐cost manner. Concludes with a description of computational experience with solving such problems.

An integrated GA---LP approach to communication network design

In this paper we demonstrate success with an implementation of a genetic algorithm, integrated with linear programming, for solving a minimum cost network synthesis problem. The problem is formulated to include a number of practical constraints and the technique applied to moderately large networks (50 nodes). The associated linear program may be large but successful methods have been developed with very small population sizes for the genetic algorithm.

The two-machine flow shop problem with arbitrary precedence relations

Abstract The scheduling problem n /3/F/ C max with arbitrary precedence constraints between the jobs is studied. A branch and bound algorithm is described. Bounds are obtained by solving 2-machine subproblems which are relaxed versions of the current 3-machine problem. These 2-machine problems which have precedence constraints can be quickly solved in an optimal fashion only if the constraints are of a series-parallel (S-P) form. Using methods described in [5], we find an S-P constraints subgraph which may be solved to obtain a lower bound for each 2-machine subproblem. This provides the basis for the calculation of an effective lower bound for the 3-machine case. Computational experience with the algorithm is reported.

A multiservice traffic allocation model for LEO satellite communication networks

Modern telecommunication networks are characterized by a heterogeneous mix of traffic classes, ranging from traditional telephone calls to video and data services. Therefore, there is a need to solve the traffic allocation problem with different quality-of-service (QoS) requirements, wherein each traffic demand is fulfilled. This paper investigates the possibility of allocating loaded traffic in a low earth orbital (LEO) satellite network by considering multiservice traffic. Both Poisson and Markov models are used for the incoming traffic, which includes multiservice requirements. By introducing a privilege parameter, a quantum of bandwidth can be reserved for high priority traffic and a better QoS can be given for this traffic. The algorithm performs well for both traffic patterns tested. Simulation results are presented.

A Branch and Bound Algorithm for Capacitated Minimum Spanning Tree Problem

This paper studies the capacitated minimum spanning tree problem (CMST), which is one of the most fundamental and significant problems in the optimal design of communication networks. CMST has a great variety of applications, such as in the design of local access networks, the design of minimum cost teleprocessing networks, the vehicle routing and so on. A solution method using branch and bound technique is introduced. Computational experiences demonstrate the algorithm’s effectiveness.

Comparison of Heuristic Search Algorithms for Single Machine Scheduling Problems

This paper compares the performance of four heuristic search algorithms for single machine scheduling problems: local search, simulated annealing, tabu search and genetic algorithms. To investigate their potential, the algorithms are applied to a single machine scheduling problem to minimise tardiness of all jobs with arbitrary ready times, processing times and due times. This problem is known to be NP complete. The purpose of the comparison is to find a good approximation algorithm ie. the algorithm is not designed to search for an optimal solution to the problem.

A node-oriented branch and bound algorithm for the capacitated minimum spanning tree problem

This paper studies the capacitated minimum spanning tree (CMST) problem, which is one of the most fundamental and significant problems in the optimal design of local computer networks. A solution method using a node-oriented branch and bound technique is introduced and its performance is presented. We show the advantages of the algorithm while illustrating the process of searching for the optimal solution. Techniques for finding tighter lower bounds are emphasized. Computational experiences demonstrate the algorithms effectiveness.

Class Dependent Traffic Allocation in a LEO Satellite Network

Traffic allocation strategy becomes a significant factor in optimization of bandwidth usage of telecommunication resources, especially with increasing use of broadband applications. Allocation strategy in dynamic LEO (Low Earth Orbital) satellite communication network is studied, to improve their Quality of Service (QoS). Traffic allocation control is performed to provide a near optimal utilization of their Inter Satellite Links (ISLs). A combination of two algorithms is used to allocate traffic in ISLs. Empirical analysis is performed to examine performance of the proposed algorithm, GALPEDA. Result shows that the proposed algorithm is useful for traffic allocation of multiclass traffic in LEO satellite communication.

A Branch and Bound Algorithm for Minimum Cost Network Flow Problem

In this paper we introduce a branch and bound algorithm for finding tree solution of minimum cost network flow problem. We consider different situations such as unit and non-unit traffic demand. The methods used to prune the searching tree in different situations are emphasized respectively but the complete searching process is not interpreted in detail due to limited space. Programming test results show the efficiency of these techniques.

A model for nonlinear cost problems in optimal design of communication networks

In this paper we present a solution strategy for nonlinear cost problems in communication network design. The cost function of the traffic on each link is modelled as piecewise constant (sequence of steps). It is shown that this model can be converted to one that describes the problem in a new way, where each single link with a complex cost function is replaced by multiple links, each with a constant cost and a capacity limit.