DSpace Collection:http://hdl.handle.net/10204/9082016-12-09T05:50:14Z2016-12-09T05:50:14ZQuantifying wave propagation over a corrugated metal using 5 dBi antennasNkosi, MCLysko, AAhttp://hdl.handle.net/10204/88322016-10-13T21:55:18Z2015-09-01T00:00:00ZTitle: Quantifying wave propagation over a corrugated metal using 5 dBi antennas
Authors: Nkosi, MC; Lysko, AA
Abstract: Understanding radio wave propagation is important for the design and implementation of reliable wireless communication systems. This paper describes transmission coefficient quantification. Measurements were done by using two antennas placed over a corrugated metal of a shipping container and also in a free space. The free space measurement is used as a reference point to study the influence of the metal on the wave propagation. The transmission coefficient measured over the shipping container is normalised to the same data measured in free space, to study the exact difference between these two scenarios. The results indicate which frequency bands are advantageous for communications over a shipping container and for which position of antenna.
Description: Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2015, Hermanus, Western Cape, 6-9 September 2015. Due to copyright restrictions, the attached PDF file only contains the abstract of the full text item. For access to the full text item, please consult the publisher's website.2015-09-01T00:00:00ZInvestigating the use of interval algebra to schedule mechanically steered multistatic radarsFocke, RWhttp://hdl.handle.net/10204/88312016-10-17T04:54:21Z2015-02-01T00:00:00ZTitle: Investigating the use of interval algebra to schedule mechanically steered multistatic radars
Authors: Focke, RW
Abstract: The findings presented in this thesis support the hypothesis that Interval Algebra (IA), as a temporal reasoning language, should perform scheduling of sensor dwells efficiently and effectively. Scheduling of multistatic radars was identified as a promising research area, as it builds upon prior research into monostatic and bistatic radars in South Africa. The hypothesis can be validated by answering three research questions. Can IA allow a multistatic radar system to make more multistatic measurements of targets? Can IA perform as well as established multisensor scheduling techniques in terms of computational requirements? Is it possible to enhance the performance of IA by making use of parallel processing architectures?
Answering the first two research questions required selecting a comparison algorithm that is already used extensively in scheduling. The Greedy Randomised Adaptive Search Procedure (GRASP) was selected as it represents two of the biggest groupings of existing scheduling algorithms. Furthermore, greedy optimisations are often preferred as they converge to optimal solutions quicker. Two scheduling scenarios were devised which made use of a binary mechanically steered surveillance radar network. One environment made use of a very simplistic model of the information fusion system; the other implemented all the details rigorously. The first environment was used to compare IA to GRASP, while the second tested a nimble IA scheduler. For both these environments Monte-Carlo simulations were used to test random target locations and motion. A novel IA algorithm that makes use of reduced point algebra was generated that allowed execution time to be reduced. Another, simpler novel contribution was an IA algorithm that ensured that radar tasks are only added to the IA network when required. Using these two techniques it was possible for IA to meet both the performance and execution time of GRASP, as allows for a richer set of constraints than required to perform multistatic scheduling. The Nimble IA Scheduler is a novel contribution which solves the realistic requirement of handling fast-moving and accelerating targets, and provides a small performance increase for the surveillance system. Answering the last research question required implementing IA on a parallel processing architecture. General-Purpose Graphical Processing Units (GP-GPUs) were selected since no published research made use these architectures and they should be well suited to solving constraint satisfaction problems. A novel parallel IA path consistency algorithm was generated in OpenCL building upon parallel versions found in the literature for supercomputers. Monte-Carlo simulations were run where both the serial and parallel versions were used to solve path consistency for randomly generated IA networks. The results for the GP-GPU identified that for large networks there was speed-up of between two to three times for consistent networks under three conditions. Firstly, the IA network must be sufficiently large to warrant copying the data to the GP-GPU. Secondly, the IA network must have a percentage of known constraints between 25% and 75%. Thirdly, the average number of IA operators should be less than 9.8. Thus, IA can provide equivalent performance to GRASP if the constraints are reduced. Given problems that require a richer set of constraints, these can easily be handled using IA. Nimble IA scheduling can provide a means to increase the multistatic measurements made and reduce those that are missed due to prediction inaccuracies. IA path consistency can also be used on GP-GPUs but only provides speed-ups under specific conditions.
Description: A thesis submitted to the Department of Electrical Engineering, University of Cape Town, in fulfilment of the requirements for the Degree of Doctor of Philosophy2015-02-01T00:00:00ZMulti-agent target tracking using particle filters enhanced with context dataClaessens, RDe Waal, ADe Villiers, PPenders, APavlin, GTuyls, Khttp://hdl.handle.net/10204/88282016-10-13T21:55:17Z2015-05-01T00:00:00ZTitle: Multi-agent target tracking using particle filters enhanced with context data
Authors: Claessens, R; De Waal, A; De Villiers, P; Penders, A; Pavlin, G; Tuyls, K
Abstract: The proposed framework for Multi-Agent Target Tracking supports i) tracking of objects and ii) search and rescue based on the fusion of very heterogeneous data. The system is based on a novel approach to fusing sensory observations, intelligence and context data (i.e. the data about the environmental conditions relevant for the tracked target). In contrast to the traditional approaches to target tracking (e.g. maritime or aviation domains), the emphasis is on tracking with low quality data sampled at low frequencies from different sensors dispersed throughout a larger area that may be only partially covered. In this demo we illustrate a live, real-time target tracking application that uses a Multi-Agent System approach to find and connect relevant information sources.
Description: Proceedings of the 2015 International Conference on Autonomous Agents and Multi-agent Systems, Istanbul, 4-8 May 2015. . Due to copyright restrictions, the attached PDF file only contains the abstract of the full text item.2015-05-01T00:00:00ZInterval algebra: an effective means of scheduling surveillance radar networksFocke, RWDe Villiers, JPInggs, MRhttp://hdl.handle.net/10204/88272016-10-13T21:55:17Z2015-05-01T00:00:00ZTitle: Interval algebra: an effective means of scheduling surveillance radar networks
Authors: Focke, RW; De Villiers, JP; Inggs, MR
Abstract: Interval Algebra provides an effective means to schedule surveillance radar networks, as it is a temporal ordering constraint language. Thus it provides a solution to a part of resource management, which is included in the revised Data Fusion Information Group model of information fusion. In this paper, the use of Interval Algebra to schedule mechanically steered radars to make multistatic measurements for selected targets of importance is shown. Interval Algebra provides a framework for incorporating a richer set of requirements, without requiring modifications to the underlying algorithms. The performance of Interval Algebra was compared to that of the Greedy Randomised Adaptive Search Procedure and the applicability of Interval Algebra to nimble scheduling was investigated using Monte-Carlo simulations of a binary radar system. The comparison was accomplished in terms of actual performance as well as in terms of computation time required. The performance of the algorithms was quantified by keeping track of the number of targets that could be measured simultaneously. It was found that nimble scheduling is important where the targets are moving fast enough to rapidly change the recognised surveillance picture during a scan. Two novel approaches for implementing Interval Algebra for scheduling surveillance radars are presented. It was found that adding targets on the fly and improving performance by incrementally growing the network is more efficient than pre-creating the full network. The second approach stemmed from constraint ordering. It was found that for simple constraint sets, the Interval Algebra relationship matrix reduces to a single vector of interval sets. The simulations revealed that an Interval Algebra algorithm that utilises both approaches can perform as well as the Greedy Randomised Adaptive Search Procedure with similar processing time requirements. Finally, it was found that nimble scheduling is not required for surveillance radar networks where ballistic and supersonic targets can be ignored. Nevertheless, Interval Algebra can easily be used to perform nimble scheduling with little modification and may be useful in scheduling the scans of multifunction radars.
Description: Copyright: 2015 Elsevier. Due to copyright restrictions, the attached PDF file only contains the abstract of the full text item. For access to the full text item, please consult the publisher's website. The definitive version of the work is published in Information Fusion, 23, pp 81-982015-05-01T00:00:00ZUncertainty representation, quantification and evaluation for data and information fusionDe Villiers, JPLaskey, KJousselme, A-LBlasch, EPavlin, GCosta, Phttp://hdl.handle.net/10204/88262016-10-13T21:55:17Z2015-07-01T00:00:00ZTitle: Uncertainty representation, quantification and evaluation for data and information fusion
Authors: De Villiers, JP; Laskey, K; Jousselme, A-L; Blasch, E; Pavlin, G; Costa, P
Abstract: Mathematical and uncertainty modelling is an important component of data fusion (the fusion of unprocessed sensor data) and information fusion (the fusion of processed or interpreted data). If uncertainties in the modelling process are not or are incorrectly accounted for, fusion processes may provide under- or overconfident results, or in some cases incorrect results. These are often owing to incorrect or invalid simplifying assumptions during the modelling process. The authors investigate the sources of uncertainty in the modelling process. In particular, four processes of abstraction are identified where uncertainty may enter the modelling process. These are isolation abstraction (where uncertainty is introduced by isolating a portion of the real world to be modelled), datum uncertainty (where uncertainty is introduced by representing real world information by a mathematical quantity), data generation abstraction (where uncertainty is introduced through a mathematical representation of the mapping between a real-world process and an observable datum), and process abstraction (where uncertainty is introduced through a mathematical representation of real world entities and processes). The uncertainties associated with these abstraction processes are characterised according to the uncertainty representation and reasoning evaluation framework (URREF) ontology. A Bayesian network information fusion use case that models the rhino poaching problem is utilised to demonstrate the taxonomies introduced in this paper.
Description: Information Fusion (Fusion), 2015 18th International Conference on Information Fusion, Washington DC, 6-9 July 2015. . Due to copyright restrictions, the attached PDF file only contains the abstract of the full text item.2015-07-01T00:00:00ZMaritime piracy situation modelling with dynamic Bayesian networksDabrowski, JJDe Villiers, JPhttp://hdl.handle.net/10204/88202016-10-13T21:55:16Z2015-05-01T00:00:00ZTitle: Maritime piracy situation modelling with dynamic Bayesian networks
Authors: Dabrowski, JJ; De Villiers, JP
Abstract: A generative model for modelling maritime vessel behaviour is proposed. The model is a novel variant of the dynamic Bayesian network (DBN). The proposed DBN is in the form of a switching linear dynamic system (SLDS) that has been extended into a larger DBN. The application of synthetic data fabrication of maritime vessel behaviour is considered. Behaviour of various vessels in a maritime piracy situation is simulated. A means to integrate information from context based external factors that influence behaviour is provided. Simulated observations of the vessels kinematic states are generated. The generated data may be used for the purpose of developing and evaluating counter-piracy methods and algorithms. A novel methodology for evaluating and optimising behavioural models such as the proposed model is presented. The log-likelihood, cross entropy, Bayes factor and the Bhattacharyya distance measures are applied for evaluation. The results demonstrate that the generative model is able to model both spatial and temporal datasets
Description: Copyright: 2015 Elsevier. Due to copyright restrictions, the attached PDF file only contains the abstract of the full text item. For access to the full text item, please consult the publisher's website. The definitive version of the work is published in Information Fusion, 23, pp 116-1302015-05-01T00:00:00Z