Upcoming Ph.D. & M.Sc. Defences


bunmi-defenceGroup peer mentorship is a relatively new addition to the area of collaborative learning. We see an untapped potential in supporting this model of mentorship with the existing collaborative learning tools like peer review and wiki. Therefore, we proposed to use a modified peer review system and a modified wiki system. From our preliminary studies using both peer review and wiki systems, we found that participants preferred the peer-review system to the wiki system in supporting them for mentorship. Therefore, this dissertation specifically addresses how to adapt the peer review system to support group peer mentorship.

We proposed a modified peer review system, which comprises seven stages – initial submission of the first draft of the paper by the author, the review of author’s paper by peer reviewers, release of review feedback to the author, back-evaluation of their reviews by the authors, modification of the paper by the author, submission of the final paper and the final stage where both authors and reviewers provide an evaluation of the peer review process with respect to their learning, their perception of the helpfulness of the process, and their satisfaction with the process. We also proposed to use our group matching algorithm, based on some constraints and the principles of the Hungarian algorithm, to achieve a diversified grouping of peers for each peer review session. With these, we conducted six peer review studies with the graduate and undergraduate students at the University of Saskatchewan and teachers in Chile. This dissertation reports on the findings from these studies.

We found that peer review, with some modifications, is a good tool to facilitate group peer mentorship. An evaluation of the performance of our group matching algorithm showed an improvement over three baseline algorithms, with respect to three metrics – knowledge gain of peers, time and space consumption. Finally, this dissertation also shows that wiki has the potential to support group peer mentorship, but needs further research.

Ph.D. Thesis Defence: Oluwabunmi Adewoyin

Wednesday, August 31, 2016

A Lightweight Attribute-Based Access Control System for IoT

samiul-defenceThe evolution of the Internet of things (IoT) has made a significant impact on our daily and professional life. Home and office automation are now even easier with the implementation of IoT. Multiple sensors are connected to monitor the production line, or to control an unmanned environment is now a reality. Sensors are now smart enough to sense an environment and also communicate over the Internet. That is why, implementing an IoT system within the production line, hospitals, office space, or at home could be beneficial as a human can interact over the Internet at any time to know the environment. 61% of International Data Corporation (IDC) surveyed organizations are actively pursuing IoT initiatives, and 6: 8% of the average IT budgets is also being allocated to IoT initiatives. However, the security risks are still unknown, and 34% of respondents pointed out that data safety is their primary concern [1].

IoT sensors are being open to the users with portable/mobile devices. These mobile devices have enough computational power and make it difficult to track down who is using the data or resources. That is why this research focuses on proposing a dynamic access control system for portable devices in IoT environment. The proposed architecture evaluates user context information from mobile devices and calculates trust value by matching with defined policies to mitigate IoT risks. The cloud application acts as a trust module or gatekeeper that provides the authorization access to READ, WRITE, and control the IoT sensor.

The goal of this thesis is to offer an access control system that is dynamic, flexible, and lightweight. This proposed access control architecture can secure IoT sensors as well as protect sensor data. A prototype of the working model of the cloud, mobile application, and sensors is developed to prove the concept and evaluated against automated generated web requests to measure the response time and performance overhead. The results show that the proposed system requires less interaction time than the state-of-the-art methods.

M.Sc. Thesis Defence: Samiul Monir

Tuesday, September 27, 2016 

Characterizing the Effects of Local Latency on Aim Performance in First-Person Shooters

zenja-defenceReal-time games such as first-person shooters (FPS) are sensitive to even small amounts of lag.  The effects of network latency have been studied, but less is known about local latency – that is, the lag caused by local sources such as input devices, displays, and the application.  While local latency is important to gamers, we do not know how it affects aiming performance and whether we can reduce its negative effects. To explore these issues, we tested local latency in a variety of real-world gaming systems and carried out a controlled study focusing on targeting and tracking activities in an FPS game with varying degrees of local latency. In addition, we tested the ability of a lag compensation technique (based on aim assistance) to mitigate the negative effects. To motivate the need for these studies, we also examined how aim in FPS differs from pointing in standard 2D tasks, showing significant differences in performance metrics. Our studies found local latencies in the real-world range from 23 to 243 ms which cause significant and substantial degradation in performance (even for latencies as low as 41 ms). The studies also showed that our compensation technique worked well, reducing the problems caused by lag in the case of targeting, and removing the problem altogether in the case of tracking. Our work shows that local latency is a real and substantial problem – but games can mitigate the problem with appropriate compensation methods.

M.Sc. Thesis Defence: Zenja Ivkovic

Friday, September 30, 2016 


colby-defenceToday’s digital games require the mastery of many different skills.  This is accomplished through play itself – sometimes experientially and other times by using explicit guidance provided by the game designer. Multiplayer games, due to their competitive nature, provide fewer opportunities for designers to guide players into mastering particular skills, and so players must learn and master skills experientially. However, when novices compete against better players – as they would if they were new to the game – they can feel overwhelmed by the skill differential. This may hinder the ability of novices to learn experientially, and more importantly, may lead to extended periods of unsatisfying play and missed social play opportunities as they struggle to improve in a competitive context. A game genre that suffers from this problem is the multiplayer first-person shooter (FPS), in which the skill difference between new players and experts who have reached a high level of expertise can be quite large. To succeed in a FPS, players must master a number of skills, the most obvious of which are navigating a complex 3D environment and targeting opponents.  To target opponents in a 3D environment, you must also be able to locate them – a skill known as opponent location awareness.  With the goal of helping novices learn the skill of opponent location aware-ness, we first conducted an experiment to determine how experts accomplish this important task in multiplayer FPS games.  After determining that an understanding of audio cues – and how to leverage them – was critical, we designed and evaluated two systems for introducing this skill of locating opponents through audio cues – an explicit stand-alone training system, and a modified game interface for embedded training.  We found that both systems improved accuracy and confidence, but that the explicit training system led to more audio cues being recognized.  Our work may help people of disparate skill be able to play together, by scaffolding novices to learn and use a strategy commonly employed by experts. 

M.Sc. Thesis Defence: Colby Johanson

Thursday, October 13, 2016