Okuthe P. Kogeda

Joint queue-perturbed and weakly coupled power control for wireless backbone networks

Abstract Wireless Backbone Networks (WBNs) equipped with Multi-Radio Multi-Channel (MRMC) configurations do experience power control problems such as the inter-channel and co-channel interference, high energy consumption at multiple queues and unscalable network connectivity. Such network problems can be conveniently modelled using the theory of queue perturbation in the multiple queue systems and also as a weak coupling in a multiple channel wireless network. Consequently, this paper proposes a queue perturbation and weakly coupled based power control approach forWBNs. The ultimate objectives are to increase energy efficiency and the overall network capacity. In order to achieve this objective, a Markov chain model is first presented to describe the behaviour of the steady state probability distribution of the queue energy and buffer states. The singular perturbation parameter is approximated from the coefficients of the Taylor series expansion of the probability distribution. The impact of such queue perturbations on the transmission probability, given some transmission power values, is also analysed. Secondly, the inter-channel interference is modelled as a weakly coupled wireless system. Thirdly, Nash differential games are applied to derive optimal power control signals for each user subject to power constraints at each node. Finally, analytical models and numerical examples show the efficacy of the proposed model in solving power control problems in WBNs.

Shilling attack detection in Collaborative Recommender Systems using a Meta Learning strategy

Collaborative Recommender Systems suggest items to a user based on other users past behaviour (items they once bought, viewed or selected and/or ratings they gave to those items). They are very effective in generating meaningful recommendations to a group of users for products or items that might interest them. However, since Collaborative filtering techniques depend on outside sources of information they are susceptible to profile injection attacks popularly known as shilling attacks. Shilling is a process in which syndicating users can connive to promote or demote a certain item. These mischievous users can consciously inject shilling profiles in an effort to bias the recommender system to their advantage. In this paper we seek to understand the degree to which shilling attacks can harm recommender systems and how these attacks can be detected. Firstly, we evaluate the vulnerabilities of collaborative filtering techniques in providing reliable recommendations. We study various attack strategies that manipulators use to attack recommender systems. Secondly we investigate the most suitable features that can be used to adequately identify shilling attacks. We propose the combiner strategy that combines multiple classifiers in an effort to detect shilling attacks. The diversity measure is used to determine the most suitable combination of classifiers. In this paper, we made use k-Nearest Neighbour, Support Vector Machines and Bayesian Networks as the initial base classifiers. The Naïve Bayes was used as a Meta Classifier. The proposed Meta-Learning classifier gave an overall performance of 99% and was found to be more superior to Neural Networks and k-Nearest Neighbor.

A review of service discovery schemes in wireless mesh networks

The increase in number of users in Wireless Mesh Networks (WMNs) setups consequently represents an upsurge in numbers of services. Services such as internet, e-commerce, audio streaming, Voice over Internet Protocol (VoIP), Video on Demand (VoD), file and printer sharing among others will be clogged and ran over WMNs. This further leads to poor Quality of Service (QoS). Quick and timely discovery of these services becomes an essential parameter in optimizing performance of these networks. In this paper therefore, we present an overview of the various existing service discovery schemes in WMNs. We also present the various gaps available in these schemes for future service discovery schemes.

Performance Optimization of Intelligent Home Networks

Home networks continue to experience an increase in the number of devices and services. This increase has come as a result of rapid technological revolutions in engineering and telecommunication industries. The advancement in technology has enabled access of intelligent home networks locally as well as remotely. This in turn has led to poor quality of service (QoS) to the consumers of such services and applications. Therefore, in this chapter, we present performance optimization of intelligent home network model that is scalable and adaptable to these increases and technological changes. We segmented and prioritized the intelligent home network into six subnets. Then we assigned weighing factor numbers to the devices, which aided in their classification and prioritization. We then grouped the supported home network services and applications into six classes and increased the number of transmitted packets per iteration in each Class of Service (CoS). We tested and evaluated proposed model using OMNET++ simulator against Priority Queuing (PQ) and Class-Based Weighted Fair Queuing (CBWFQ) models. The results show an average network packet throughput of 99.74 %, delay of 3.02 s, and loss of 1.59 %.

A cloud computing augmenting agricultural activities in Marginalized Rural Areas: A preliminary study

Marginalized Rural Areas (MRAs) practice farming and in most cases lack basic resources and skills to improve yields, which are often poor. This has led to famine, poverty, crime and rural to urban migration. Agricultural activities when practiced very well can alleviate such challenges in the country. We are therefore, developing a cloud computing model that seeks to improve agriculture as an activity in MRAs. It shall include a cloud architecture which is a Mobile information system and would be used by farmers (including subsistence farmers) to share (upload and download) information about the farming techniques, markets, weather, seeds, etc. In this paper, we present findings of a preliminary study of what common mobile devices are mostly used and what type of agriculture is mostly practiced by MRAs people. We collected data through close ended questionnaires that were given to farmers in Pongola in KwaZulu-Natal province of South Africa. The preliminary study results show that 54% of farmers face challenges of pests and birds, 26% weather and 20% diseases affecting their livestock. 94% of farmers own mobile devices of which 59% of them are feature phones. 58% of the mobile phone owning farmers do not use them to access information regarding farming and lastly the most practiced type of agriculture in MRAs is subsistence farming with 97%, where famers focus on growing enough food to feed their own consumption.

A Model of Controlling Utilization of Social Grants in South Africa

South Africa is one of the most unequal countries in the World. Social grant in South Africa is supported by the Social development macro policy framework since 1994. It was aimed at poverty alleviation that combines social and economic goals. Government over the years has been faced with a number of challenges in social grants and benefits administration such as fragmented institutional arrangements and a lack of uniformity, fraud and corruption. Despite the fact that the government took steps in minimizing fraud and corruption associated with the social grants administration, the challenges of utilization of funds disbursed still exist. This research seeks to design and implement a system that controls utilization of social grants using Near Field Communication technology. In this paper, we present the system design and architecture of the system to be implemented. We present the findings of a preliminary study of the utilization of social grants by recipients. We collected data by interviewing social grant recipients in South Africa. The results of this preliminary study shows that 62% of social grant recipients use their funds within a week, 82% within two weeks, and 92% within three weeks. Only 8% use the funds until the fourth week.

Modeling of Service Discovery Over Wireless Mesh Networks

Wireless Mesh Networks (WMNs) have played a huge rule in networking environments by supporting seamless connectivity, Wide Area Networks (WANs) coverage, mobility features, etc. However, the rapid increase in the number of consumers on these networks brought an upsurge in competitions for available services and resources. This has led to link congestions, data collisions, and link interferences, which affects Quality of Service (QoS) . Therefore, the quick and timely discovery of the services and resources becomes an essential parameter in optimizing the performance of service discovery on these networks. In this study, we present Ndlovu Okuthe Manoj (NOM) model, a service discovery model that integrates the Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) algorithms. The PSO is used to dynamically define and give different priorities to services on the network, based on varied workflow procedures. On the other hand, the ACO is used to effectively establish the most cost-effective path whenever each transmitter has to be searched to identify whether it possesses the requested service(s). Furthermore, we design and implement the Link Collision Reduction (LCR) algorithm. It’s objective is to define the number of service receivers to be given access to the services simultaneously. We then simulate the proposed model in Network Simulator 2 (NS2), against Ant Colony based-multi constraints QoS-aware service selection (QSS) and FLEXIble Mesh Service Discovery (FLEXI-MSD) models. The results show an average service discovery throughput of 80%, service availability of 96%, service discovery delay of 1.8 s, and success probability of service selection of 89%.

Towards a Model of Enhancing Safety Fishing in South Africa

In South Africa, fishing industry helps reduce poverty, famine and crime by providing jobs to citizens. Fishing industry also contributes to income generation of the country through export markets, investments from private companies, etc. However, the industry faces a mirage of challenges due to unsafe fishing conditions caused by weather, criminals, wild animals, faulty boats, etc. In other words, the benefits we get from fishing are affected by these unsafe conditions. We intend to implement a system that would enhance safe fishing by reducing unsafe fishing conditions using Java, Android, MySQL, and Toad technologies. In order to model our system, we went to Eastern Cape, Limpopo, Mpumalanga, Western Cape, Northern Cape and Kwazulu Natal provinces to collect data using questionnaires. Research questions included, what methods fishermen use to catch fish, what challenges fishermen face while fishing, etc. After collecting all the required data, we modelled the system using UML diagrams. The results show real challenges and safety concerns for fishermen in South Africa.

A Model Augmenting Credit Risk Management in the Banking Industry

A lack of reliable credit risk measurements and poor control of credit risks has caused massive financial losses across a wide spectrum of business. Financial institutions like banks have not been able to control and contain the rapid increases of the credit defaulting. In this paper, we address the credit lending challenges by eliminating credit defaulting faced by the banking industry. Data from bank of previously accepted and rejected loan applicants was used to construct a credit risk evaluation network. The artificial neural network technique with back-propagation algorithm was applied to develop a model that supports the banks in the credit granting decision-making. The model was trained to categorize applicants as either good credit granted or bad credit denied based on the credit record. The model was able to predict whether a particular applicant is likely or unlikely to repay the credit. The training of neural network model and validation testing was done using data obtained from the bank. The results show a greater performance, classification and prediction accuracy.

An Improved-Cross Layer Scheduling model for intelligent home networks

We daily add more devices and services into existing intelligent home networks. Consequently, various networking standards evolutions being experienced world over have not left home networks behind. These evolutions results in competition and depletion of the available limited resources. Consumers therefore, experience unavailable, unreliable and poor performing network both locally and remotely. In this paper, we present an Improved-Cross Layer Scheduling (CLS) model that optimizes performance of intelligent home networks. We have used Particle Swam Optimization (PSO) algorithm to dynamically schedule, align and prioritize network subnets, devices and services in the model. We have used Virtual Local Area Network (VLAN) protocol to classify home network into six subnets. Consequently, we have used weighing factor numbers to classify subnet devices. Further, we have used Differentiated Service Code Points (DSCP) to classify supported home network services into six classes. Moreover, we have increased number of packets transmitted per Class of Service (CoS) iteratively. We have optimized each subnet and used the output of the preceding subnet as input to subsequent subnet. Equally, we have reduced delay between consecutive transmitting CoSs in the media. We have simulated our model and realized average network throughput of 99.735%, packet loss of 1.59% and delay of 1.82 milliseconds.