South African Computer Journal - Volume 2005, Issue 34, 2005
Volumes & issues
Volume 2005, Issue 34, 2005
Author Derrick G. KourieSource: South African Computer Journal 2005 (2005)More Less
SACJ's position with respect to conference publications should be made explicit. The fact that the Department of Education's subsidy policy discriminates against conference contributions, means that there is a risk that local authors might withhold their best work from conferences, aspiring instead to obtain a journal publication.
Hybrid traffic engineering : from Constraint Shortest Path First to Least Path Interference : reviewed articleAuthor Antoine B. BagulaSource: South African Computer Journal 2005, pp 2 –10 (2005)More Less
This paper presents a new approach for routing flows in IP networks. The approach referred to as the Least Path Interference (LPI) is based on a route optimisation model which (1) moves the traffic away from path interfering links (the path interference quantifying the network reliability) to re-route fewer flows upon link failure and (2) maximises the link congestion distance (quantifying the network optimality) to reject fewer flows under congestion. LPI implements a hybrid traffic engineering model combining offline estimation of the path interference and online path selection. LPI is based on a simple path selection model where no changes to the traditional routing algorithms are required besides designing a new mixed cost metric to combine reliability and optimality. The Least Path Interfering Algorithm (LPIA ); a routing algorithm derived from LPI is applied to compute paths for the traffic offered to a 20- and 50-node networks. Simulation reveals (1) performance improvements compared to Open Shortest Path First (OSPF) and Constraint Shortest Path First (CSPF) routing in terms of routing optimality and network reliability and (2) the same performance as the recently proposed Least Interference Optimisation Algorithm (LIOA) algorithm with less signalling overheads.
Source: South African Computer Journal 2005, pp 11 –25 (2005)More Less
Regular tree grammars and top-down tree transducers are extended by random context sensitivity as known from the areas of string and picture generation. First results regarding the generative power of the resulting devices are presented. In particular, we investigate the path languages of random context tree languages.
Author M.G.G. LaidlawSource: South African Computer Journal 2005, pp 26 –32 (2005)More Less
Run-length encoding is used in applications that process digitised analogue data. Lossless image compression is an example. Golomb encoding is optimal for infinite geometric probability distributions. A wider class of codes is presented here, all of which are suitable for encoding the elements from an infinite set. It is shown that each code can be described by a polynomial k(L), which determines the number of codewords of lengthL. An even wider class of codes is also considered in which the number of codewords grows geometrically with their length. Criteria are given for excluding such codes from further consideration.
Author Martin S. OlivierSource: South African Computer Journal 2005, pp 33 –40 (2005)More Less
This paper introduces a Privacy-Enhancing Technology (PET) based on a hybrid of Crowds and anonymising proxies. The PET - referred to as Flocks - operates by establishing a number of Web proxies and letting these proxies randomly forward requests to other proxies (or the final destination). This distributes users' requests over a number of such proxies, thereby helping to protect their (browsing) privacy.
The problem that the paper considers is the effect of two primary design parameters on the privacy of the overall system. These parameters are the probability with which a proxy sends a request to the destination server rather than another proxy (α) and the number of proxies in the system (N). Two privacy objectives are identified, namely the number of hops used to satisfy a request and the portion of proxies that 'know' about a request. A third requirement deals with the external communication costs of the system. A formal analysis is performed to determine these three factors from the two identified parameters. Finally numerical examples are used to explore the impact of these two parameter choices in concrete terms.
The proposed PET differs from existing PETs in two significant manners: It is primarily intended to be used inside an organisation to protect the privacy of users inside the organisation (in particular, employees) and it takes explicit cognisance of forensic factors.
Author S.T. RockSource: South African Computer Journal 2005, pp 41 –51 (2005)More Less
Focussing on the temporal structures of the events described by language, the sentence 'Simmer the soup stirring occasionally for fifteen minutes'
is ambiguous between five semantic interpretations. Some of these readings can only be found if we take the view, argued by Moltmann, that temporal measure adverbials play the role of part quantifiers, rather than the more traditional approach of treating them as predicates.
In this paper I describe first of all the justification for adopting the quantifier view of measure adverbials, and argue for its extension in including object quantifiers. I then demonstrate how this is used in a computational system to expose some ambiguity that would otherwise remain hidden.
Source: South African Computer Journal 2005, pp 52 –68 (2005)More Less
In this paper we report research conducted into program test case generation using Genetic Algorithms (GA). Our goal was to produce an effective test case generation technique for programs which will at the same time be simple enough as a tool for teaching software engineering courses. In the proposed approach, we combined information about program input domain with knowledge of the structure of the program under test in order to generate optimal test cases. Features of the test elements including domain attributes, test case selection criteria, and test values were coded as genes in the chromosome used to represent test cases. Our approach allows the automation of many aspects of test case generation including program instrumentation and test case evaluation. The implementation framework is described and the prototype system is illustrated using Ada programs for linear and binary search. Although the case studies for testing the prototype system are not as complex as many real-life problems, results of the experiments indicate that our approach is practical, effective and capable of generating adequate test-cases for program testing while at the same time simple enough to be useful for the purpose of demonstrating software testing process to software engineering students. Current work is focused on improving the system by automating several aspects of program instrumentation and building a user interface for computer assisted teaching.
Source: South African Computer Journal 2005, pp 69 –75 (2005)More Less
We investigate the minimization of unary symmetric difference nondeterministic finite automata. To this end, we develop a modification of the Berlekamp-Massey algorithm. We then prove that, given a unary symmetric difference nondeterministic finite automaton with one final state, this modified algorithm produces a minimal symmetric difference nondeterministic finite automaton which accepts the same language. We also discuss the application of the technique in the case where the nondeterministic finite automaton has more than one final state. Finally, we consider compression, as opposed to minimization, of symmetric difference nondeterministic finite automata.
Source: South African Computer Journal 2005, pp 76 –83 (2005)More Less
Nearly all present-day commercial intrusion detection systems are based on a hierarchical architecture. Nodes at the bottom of the hierarchy collect information, which is passed to higher nodes in the hierarchy until the root node is reached. The root node is a command and control system that is responsible for detecting intrusions and for issuing responses. However, an intrusion detection system (IDS) based on a hierarchical architecture has many single points of failure. For example, by disabling the root node, the intrusion-detection function of the IDS will also be disabled. To solve this problem, we propose an IDS inspired by the human immune system. The proposed IDS has no single component that is responsible for detecting intrusions. Instead, the intrusion-detection function is divided and placed within mobile agents. Mobile agents act similarly to white blood cells of the human immune system and travel from host to host in the network to detect intrusions. The proposed IDS is fault-tolerant because it can continue to detect intrusions even when most of its components have been disabled. Furthermore, because mobile agents are not static and their number can vary, the whole IDS is more difficult to disable than an IDS based only on static components.
Author Stefan GrunerSource: South African Computer Journal 2005, pp 85 –86 (2005)More Less
For the readers of the South African Computer Journal it will be interesting to know that Derrick Kourie, editor of this journal, belonged to the international Programme Committee of SAC-SE-05, making great contributions to technical quality with his thoughtful paper reviews.
Evolving intelligent game-playing agents, South African Computer Journal, 32 Jun 2004 : pp. 44-49 : erratumSource: South African Computer Journal 2005 (2005)More Less
Joint special issue of ARIMA and SACJ, on "Advances in end- user data mining techniques" : call for papersSource: South African Computer Journal 2005, pp 88 –89 (2005)More Less