Academia and Publications


If you want to book a meeting with me, get in touch by eMail. My Teaching calendar is shown below.

Publications and Presentations

The sum of my contributions to Computer Science over the years, including my original two dissertations.

Author in…

Dealing With Big Data Outside Of The Cloud: GPU Accelerated Sort
The demands placed on systems to analyse corpus data increase with input size, and the traditional approaches to processing this data are increasingly having impractical run-times. We show that the use of desktop GPUs presents a significant opportunity to accelerate a number of stages in the normal corpus analysis pipeline. This paper contains our exploratory work and findings into applying high-performance computing technology and methods to the problem of sorting large numbers of concordance lines.
PDF Download

Estimating node lifetime in interference environments
For commercial Wireless Sensor Network (WSNs) deployments it is necessary to estimate the network lifetime. It must be possible before network deployment to determine how long a network maintains operational before maintenance is required and batteries have to be replaced. Unfortunately, node lifetime is very dependent on the radio environment in which the node is operated. As we will demonstrate in this paper the node lifetime in a very busy radio environment can be up to 11 times shorter than in a quiet environment. WSNs employ duty-cycled communication protocols where receivers periodically sample the channel to determine if it has to remain active to receive a message. Radio interference triggers the receive mechanism causing an unnecessary wake-up which leads to an increase in a node’s energy consumption. In this paper we present a method for estimating node energy consumption in a target radio environment. We describe how to capture the essential characteristics of the radio environment and how to use this information to predict node lifetime. We demonstrate the usability of the proposed method using the well known WSN communication protocol ContikiMAC. Our evaluation comprising real-world scenarios shows that the proposed method is able to accurately predict node lifetime.

ShapeClip: Towards Rapid Prototyping with Shape Changing Displays for Designers
This paper presents ShapeClip: a modular tool capable of transforming any computer screen into a z-actuating shape-changing display. This enables designers to produce dynamic physical forms by ‘clipping’ actuators onto screens. ShapeClip displays are portable, scalable, fault-tolerant, and support runtime re-arrangement. Users are not required to have knowledge of electronics or programming, and can develop motion designs with presentation software, image editors, or web-technologies. To evaluate ShapeClip we carried out a full-day workshop with expert designers. Participants were asked to generate shape-changing designs and then construct them using ShapeClip. ShapeClip enabled participants to rapidly and successfully transform their ideas into functional systems.
PDF Download

Poster: Resilience of the Internet: the Case of the BGP Backbone
The Internet has become a critical infrastructure. This multi-disciplinary study evaluates the resilience of one of its cornerstone assets: the Border Gateway Protocol (BGP) infrastructure, that constitutes the backbone of the network of networks. The study combines three complementary approaches: a Systematic Literature Review (SLR) established an exhaustive threat model for BGP systems; a qualitative analysis, based on a series of interviews with major stakeholders of the Internet, identified key specificities of the operation and governance of backbone networks; and large scale attacks were simulated based on a detailed study of the logical and physical topologies of the backbone. Our first conclusions challenge the idea that the Internet’s built-in resilience guarantees against large-scale failures: important changes in the way the backbone is operated and governed may be necessary to address new threats arising in a near future.
PDF Download

RelyOnIt Project
Existing IoT solutions do not provide dependable performance. A major reason for this is that embedded wireless sensors and actuators are deeply affected by their often hostile environment. Radio interference from other wireless equipment and electrical appliances impairs communication, temperature and humidity variations affect battery capacity and electronics.

RELYonIT is currently trying to close this gap by providing a systematic framework and toolchain to enable dependable IoT applications by taking into account all relevant environmental properties and their impact on IoT platforms and protocols. Environment-aware IoT protocols will be developed and automatically configured to meet application-specific dependability requirements.

  • D-1.3 Report on Runtime Assurance – PDF Download
  • D-4.4 Final Integrated Prototype and Experiment – PDF Download
  • Report on Protocol Models & Validation and Verification – PDF Download


Acknowledged in…

Exploring Interactions with Physically Dynamic Bar Charts
Visualizations such as bar charts help users reason about data, but are mostly screen-based, rarely physical, and almost never physical and dynamic. This paper investigates the role of physically dynamic bar charts and evaluates new interactions for exploring and working with datasets rendered in dynamic physical form. To facilitate our exploration we constructed a 10×10 interactive bar chart and designed interactions that supported fundamental visualisation tasks, specifically: annotation, navigation, filtering, comparison, organization, and sorting. The interactions were evaluated in a user study with 17 participants. We identify the preferred methods of working with the data for each task (e.g. directly tapping rows to hide bars), highlight the strengths and limitations of working with physical data, and discuss the challenges of integrating the proposed interactions together into a larger data exploration system. In general, physical interactions were intuitive, informative, and enjoyable, paving the way for new explorations in physical data visualizations.
PDF Download | ACM Listing


This Post Has Been Viewed 650 Times