Toni Anwar

A genetic algorithm-based support vector machine for bus travel time prediction

Information about public transport travel time is a key indicator of service performance, and is valued by passengers and operators. Among many different approaches, Support Vector Machines (SVM) has recently gained attention in predicting bus travel times. The training process of SVMs involves solving a quadratic programming problem which is slow when dealing with large training data. This paper proposes a Least Squares SVM (LS-SVM) method that expedites the training process by simplifying the quadratic programming problem using a linear regression technique. Also, to ensure the accuracy of the prediction results, a Genetic Algorithm (GA) is used to determine the optimal set of model parameters. The GA based LS-SVM approach is tested using real-world travel time data from a bus route in Melbourne, Australia. The comparison of the results in this paper to those obtained in a previous study using artificial neural networks shows that the proposed method produces more accurate results.

A Novel Quantum-Dot Cellular Automata XOR Design

Quantum-dot cellular automata (QCA) is an emerging nanotechnology that promises faster speed, smaller size, and lower power consumption compared to the transistor-based technology. Moreover, XOR is a useful component for the design of many logical and functional circuits. This paper proposes a novel and efficient QCA XOR design. The proposed XOR design has been compared to a few recent designs in terms of area, speed and complexity. Comparison of results illustrates significant improvements in our design as compared to traditional approaches. Also simulation proves that the proposed XOR design is completely robust and more sustainable to high input frequency as compared to other designs. This robustness is highly significant when this component is applied for realizing larger designs.

Comparison binary search and linear algorithm for German-Indonesian sign language using Markov model

Research in the area of sign language gives significant results in the decade of years. A lot of research has been done focusing on the signer-independent schemas that contain sign language and video. This paper contributes the statistical translation words from German-Indonesia sign language and vice versa. The process of translation used Markov model and parsing tree to translate words. A binary search and linear algorithms in collaboration with Markov model are used to implement the translation process. The execution time between 2 models of algorithms has been analyzed. A binary search and Markov model have given a better execution than sequential algorithm and Markov model when translates words.

Analysis of Adaptive Policy-Based Approach to Avoid Policy Conflicts

PobMC is an adaptive scalable framework which uses policies to control and adapt the system behaviour. Moreover, PobMC has the capability to decouple the adaptation concerns from the application code. Since policies are used to govern the system behavior, conflicts may arise in the set of policies and also may arise during the refinement process, between the high-level goals and the implementable policies. Moreover, policy conflict can result from propagation, action composition and other constraint policies, which cannot be detected by simply comparing authorization policies. In this paper we present a static analysis to address the inconsistencies, scalability when there are two or more policies are enforced simultaneously. Then, we classify our system policy conflicts to verify that policies are enforced correctly. Moreover, the paper provides temporal specification patterns to detect each type of conflicts.

RoadCrowd: An approach to road traffic forecasting at junctions using crowd-sourcing and Bayesian model

Real time road traffic update is highly desirable for city dwellers, commuters, tourists and travelers. In this paper, we propose a novel methodology for estimating traffic conditions at road intersections. Our methodology involves crowd sourcing approach for traffic estimation as well as utilization of conditional probability distribution of traffic states at adjacent junctions. The later approach is inspired by Bayesian inference. Bayesian inference is used to estimate traffic condition at one junction based on the traffic condition at an adjacent junction. We have formulated two algorithms which are used to update traffic conditions and to retrieve traffic conditions respectively based on the crowd sourced data and the probabilistic methodology. A web based prototype has been developed which incorporates the described methodology. The initial evaluation shows the feasibility of our proposed methodology.

The MedMaps apps: Mobile application for finding, managing and commercialize pharmacy

Nowadays, there are many inventions of technology, websites and mobile applications that are able to solve real-life problems people were once facing. The scope for this development of mobile application is focusing on medical field in which the purpose is to help people looking for medicine in the nearest pharmacy from their current location. The process of finding pharmacy is not as easy as finding any other shops, especially to locate the pharmacy that is selling the exact medical product that users are looking for. The consequences will be a time consuming and waste of money if they had to drive all away to the wrong pharmacy. That is how the idea of this MedMaps: Mobile Application for Finding, Managing and Commercialize Pharmacy is generated from. This application also provides Pharmacy a platform to commercialize their store and promoting their products. As for public users, they can also manage their pill consumption intake by setting an alarm that will always remind them when to consume the tablets. The development of this application has going through all required phases based on the Rational Unified Process (RUP) software methodology. The analysis had been done and requirements are being documented properly. To ensure the efficiency of the end product, the literature review on existing system and technology are also important. In addition, the decision on the chosen software architecture plays an important role in organizing requirements, maintaining the quality and help supporting this application development. Other than software architecture and the results from requirements elicitation, low-fidelity prototype produced also helps in supporting the construction of this mobile application.

Analysis of motivation approach in mobile crowdsensing application: Specialize on public transportation domain

Mobile Crowdsensing s a new sensing paradigms which bring the revolutionary advance in crowdsourcing application. Although there are many possibilities in crowd sensing application, motivation and incentive is a critical issue when the devices have very limited resources (e.g., energy and storage capacity) or the information revealed is highly sensitive. In this research, we design a motivation model in public transport domain through identify the key component that need motivation, designing Android applications of the model and evaluating the applications with real life data. The suggested model helps to provide a low budget to motivate user in government or non-profit organization Mobile Crowd Sensing application and maintain long lasting contributor in their application.

Context Aware Handover Algorithms for Mobile Positioning Systems

Abstract: This work proposes context aware handover algorithms for mobile positioning systems. The algorithms perform handover among positioning systems based on important contextual factors related to position determination with efficient use of battery. The proposed solution which consists of the algorithms is implemented in the form of an Android application named Locate@nav6. The performance of the proposed solution was tested in selected experimental areas. The handover performance was compared with other existing location applications. The proposed solution performed correct handover among positioning systems in 95% of cases studied while two other applications performed correct handover in only 50% of cases studied. Battery usage of the proposed solution is less than one third of the battery usage of two other applications. The analysis of the positioning error of the applications demonstrated that, the proposed solution is able to reduce positioning error indirectly by handing over the task of positioning to an appropriate positioning system. This kept the average error of positioning below 42.1 meters for Locate@nav6 while the average error for two other applications namely Google Latitude and Malaysia maps was between 92.7 and 171.13 meters.