Peter J. B. King

Derivation of Petri Net Performance Models from UML Specifications of Communications Software

As hardware becomes faster and bandwidth greater, the determination of the performance of software based systems during design, known as Software Performance Engineering (SPE), is a growing concern. A recent seminar of experts at Dagstuhl and the First International Workshop on Software and Performance have both highlighted the need to bring performance evaluation into the software design process. The Unified Modelling Language (UML) has emerged in the last two years as a widely accepted standard notation for software design and it is an attractive vehicle for SPE. In this paper UMLs Collaboration and Statechart diagrams are shown to allow systematic generation of Generalised Stochastic Petri Net (GSPN) models, which can be solved to find their throughput and other performance measures. Using the example of communication via the alternating bit protocol, such a mapping is demonstrated and the resulting GSPN solved using the SPNP package. The basis of a usable methodology for SPE is explored.

Nested algorithmic skeletons from higher-order functions

Algorithmic skeletons provide a promising basis for the automatic utilisation of parallelism at sites of higher order function use through static program analysis. However, decisions about whether or not to realise particular higher order function instances as skeletons must be based on information about processing resources available at runtime In principle, nested higher order functions may be realised as nested skeletons. However, where higher order function arguments result from partially applied functions, free-variable bindings must be identified and communicated through the corresponding skeleton hierarchy to where those arguments are actually applied Here, a skeleton based parallelising compiler for Standard ML is presented. Hybrid skeletons, which can change from parallel to serial evaluation at runtime, are considered and mechanisms for their nesting are discussed. The main compilation stages are illustrated for simple examples. A nested higher order function based algorithm for multiplying matrices of arbitrary length integers is presented along with performance figures for compiled code running on a Fujitsu AP3000.

Analysis of security attacks on AODV routing

MANET routing protocols have many vulnerabilities that may be exploited by malicious nodes to disrupt the normal routing behavior. In this paper, we present a vulnerability analysis of AODV. We simulate four routing attacks to analyse their impacts on AODV protocol using NS-2 network simulator. These attacks are blackhole, grayhole, selfish and flooding attacks. The blackhole and flooding attacks have a severe impact on the network performance while the selfish and grayhole attacks have less significant effect on the network performance.

Two approaches to integrating UML and performance models

R.P. Hopkins Department of Computing and Electrical Engineering, Heriot-Watt University, Edinburgh EH14 4AS, Scotland rph@cee.hw.ac.uk M.J. Smith Department of Computing and Electrical Engineering, Heriot-Watt University, Edinburgh EH14 4AS, Scotland ceemjs@cee.hw.ac.uk P.J.B. King Department of Computing and Electrical Engineering, Heriot-Watt University, Edinburgh EH14 4AS, Scotland pjbk@cee.hw.ac.uk

Experiences With Online Programming Examinations

An online programming examination was used to assess undergraduates who had learned Java for two terms. The advantages are that the students are assessed on what they have to do in practice, rather than on their theoretical knowledge, and that the marking load on the examiner is drastically reduced. The style of question used and the results of the examination are analyzed.

Performance Evaluation of Routing Protocols for MANET

An ad hoc network is a collection of autonomous mobile nodes to forming an instant multihop radio infrastructure-less network in a dynamic topology. Each node in Ad hoc network, functions as host and router at the same time,where maintaining connectivity in a decentralized manner by distributing the network control among the nodes. A large number of routing protocols for mobile ad-hoc networks are presented since last decade. Topology frequently changing,transmission power and asymmetric links are the main challenge that type of protocols is facing. Therefore, both proactive and reactive routing protocols for mobile ad-hoc networks show incapability, and sometimes prove to be inefficient under these circumstances. We have previously developed and presented a new multipath hybrid routing protocol in name of Multipath Distance Vector Zone Routing Protocol and we refer to it as MDVZRP. The protocol is implemented in the well known network simulator (ns2). In MDVZRP we assume that all the routes in the routing table are active and usable at any time, unless the node received or discovered a broken link. Therefore, no need to update the routing tables periodically to reduce the periodic update messages and hence we reduce the control traffic in the entire network. The protocol guarantees loop freedom, disjoint paths,and alternative paths in case of break the active path by none frequently maintaining topological map of a zone centered on each node. In this paper, the MDVZRP’s performance is evaluated and compared with DSDV and AODV standard protocols delivering CBR traffic. Simulation results show that it gives better performance than DSDV in all situations, and better than AODV, when mobility is low.

Approximate Estimation of Transaction Response Time

This paper describes a study of different approximation techniques used to predict the response times of database transactions represented as patterns of resource consumption and modelled with non-product-form queueing networks. The techniques are applied to a range of examples. The experiments show that none of the approximation techniques has a consistent advantage over the others for all cases considered. On the other hand, a simple heuristic rule is formulated which provides an acceptable approximation to the average transaction response time for the entire range of examples. The rule specifies a procedure for labelling each queue in a queueing network as either an M/M/1 or an M/G/1 resource. The resulting network can then be solved to obtain the mean response time of individual transactions.

Experiences with group projects in software engineering

Computer Science graduates are often accused of having no experience of the real world problems of large software systems. In general, the practical work is said to be too small scale, and too individually oriented. This paper describes the introduction of a major group project into the degree syllabus at Heriot-Watt University. The project is designed to give the students experience of the problems involved in writing multi-author software and meeting strict deadlines. The choice of topics, supervision and management aspects are all covered. The problems of assessment are described.

Numerical methods for infinite Markov processes

The estimation of steady state probability distributions of discrete Markov processes with infinite state spaces by numerical methods is investigated. The aim is to find a method applicable to a wide class of problems with a minimum of prior analysis. A general method of numbering discrete states in infinite domains is developed and used to map the discrete state spaces of Markov processes into the positive integers, for the purpose of applying standard numerical techniques. A method based on a little used theoretical result is proposed and is compared with two other algorithms previously used for finite state space Markov processes.

An intermittently served discrete time queue with applications to meteor scatter communications

A discrete time single server queue with service interruptions is analyzed in the steady-state under general assumptions. The main motivation for the study is the performance evaluation of a communication protocol using ionized layers created by meteors. The analysis yields the joint distribution of the queue size and the remaining duration of the current operative or inoperative period. The solution takes a particularly simple form in the case where the operative periods have a rational generating function.