Dr Leandro Soares Indrusiak
Lecturer in Real-Time Embedded Systems
|
|
Dr Leandro Soares Indrusiak |
Open Projects 2012(to be taken in 2012-2013)Leandro Soares Indrusiak |
|
Requirements: creativity and self-motivation Desired: good background in algorithms, good programming skills (specially if the interface with auction websites is to be built), solid background in quantitative analysis Suitable for: CS, CS/Maths, MEng, MScNC, MScSWE, MScComp, MScIT More information: 12.lsi.2 - Scheduling topology reconfiguration in Networks-on-ChipThis project assumes a network-on-chip architecture that can dynamically reconfigure its topology. Given a particular application, the goal of the project is to devise an algorithm that can schedule reconfigurations in such a way that it minimises the worst case behaviour of the application (i.e. by creating shortcuts, duplicating congested links). The trade-off to be explored has to take into account the following aspects:
A successful project should produce a scheduling algorithm and an evaluation infrastructure (i.e. simulation models or static analysis) that can comparatively evaluate different schedules. Requirements: strong background in computer architectures, embedded systems Desired: networks-on-chip (EDI), response time analysis (RTS) Suitable for: CS, CS Emb, MEng More information: 12.lsi.3 - Heuristics for dynamic mapping of tasks in multicore systemsThe goal of this work is to implement and compare dynamic mapping heuristics, so that application tasks can be assigned and/or migrated to different processors during runtime in a multicore system. Important points to be observed when implementing the mapping heuristics include:
Requirements: good background in algorithms, concurrent programming Desired: scheduling algorithms (RTS / CRT) Suitable for: CS, CSEmb, MEng More information: 12.lsi.4 - Abstract models for estimating energy consumption in multicoresThe goal of this work is to model energy consumption into a simulation framework for multicores. It should focus mainly on the energy dissipated by the on-chip interconnect structure (e.g. on-chip bus, network-on-chip), and should include both dynamic and static power consumption. The outcome of the project should be a graphical interface that will show the power dissipation in different parts of the chip as the simulation executes.Different modelling approaches can be used, and the best solution is probably a mixture of the following:
Requirements: embedded systems design (EDI) Suitable for: CSEmb, MEng More information: 12.lsi.5 - Benchmarks for multi- and manycore systemsThis project involves the identification, customisation and use of benchmarks for multi- and manycore systems. Benchmarks have always been used to compare how well different architectures perform under specific conditions. However, as systems get more complex a new class of benchmarks is needed, which are able to evaluate systems at a higher level of abstraction (i.e. higher than processor instruction level). In this project, the student will have to review existing uniprocessor and multiprocessor benchmarks such as SPEC and PARSEC and do literature survey to check how such benchmarks have been used to validate large multi- and manycores. The final part of the project is the investigation of a suitable approach to use such benchmarks to evaluate end-to-end latencies (computation and communication) using transaction-level simulation models of large multicores.A successful project should produce a good literature survey on benchmarks for multicores and adapt at least one of them so that it can be used to drive a transaction-level simulator of a large multicore. Requirements: strong background in computer architectures, embedded systems Desired: multiprocessors and networks-on-chip (EDI) Suitable for: CS, CS Emb, MEng More information: 12.lsi.6 - Simulation of video processing algorithms using Ptolemy IIThis project will take a simple video processing algorithm and integrate it into Ptolemy II simulator, so that the performance of the algorithm can be evaluated under different levels of parallelism (i.e. single-core, dual-core with shared memory, multi-core with distributed memory). The simulation should actually include both functional and non-functional parts of the system (i.e. the actual processing of the video, as well as the timing overheads of each part of the algorithm).The initial choice of the algorithm is a Sobel filter for edge detection, but a different choice can be proposed by the student. Implementations of this algorithm is available within the Silhouette Project, and this could be used as a starting point. A successful project will allow the simulation of the processing of a video stream (from a file or camera) and the monitoring of timing metrics for different parallel architectures. Requirements: excellent Java programming skills Desirable: embedded systems design (EDI), image processing (VIGR) Suitable for: CS, CSEmb, MEng More information: 12.lsi.7 - Evaluation of the impact of cycling in urban areasThis open-ended project aims to evaluate the impact of encouraging cycling in urban areas as a way to reduce society's carbon footprint. It should use simulation models to compare different scenarios where different proportions of the population uses bicycles or individual cars. The comparison metrics should include carbon emissions and time to reach the destination.The evaluation should consider an area of York and have an accurate description of the road network for that chosen area (e.g. obtained from OpenStreetMap.org). Interesting issues to be considered include:
A successful project will enable the analysis and simulation of different scenarios with different balance of cyclists/drivers, and provide metrics for carbon emissions and time to destination. Part of the project should also be a discussion of the relevance of the obtained results, and the ethical issues about using this kind of evaluation to guide and influence public policies. Requirements: good programming skills, creativity Suitable for: CS, MEng, MScSWE, MScComp, MScIT Related work: 12.lsi.8 - Evaluating the correlation of real-world and online social networksThis open-ended project aims to evaluate social networks in the real world, and correlate them with their online counterparts. The evaluation system will be composed by three modules:
A number of extensions can be considered, such as considering the time and place of interactions in the real world (e.g. to investigate differences between work-related and leisure-related social network). A successful project will have a prototype implementation of the three modules above and will perform a number of small scale experiments with 10-20 users. A similar project using MoteRunner is also available: 12.lsi.9. Requirements: good programming skills, creativity Desired: experience with app programming (Android preferred) Suitable for: CS, MEng, MSc HCIT, MScSWE, MScComp, MScIT Related work: Background information for Ambient Intelligence projects based on IRIS motes and MoteRunner: 12.lsi.9, 12.lsi.10 and 12.lsi.11The following three projects will be based on Mote Runner, which is both a programming environment and a runtime platform for wireless sensor network motes. It supports Java and C#, and is fully integrated to Eclipse, so applications can be developed, debugged and simulated within a high-level programming and simulation environment before they are uploaded to the actual motes.All projects are expected to develop prototypes based on the IRIS motes available at the department. A total of 30 motes and 10 interface boards will be made available. More information: 12.lsi.9 - Monitoring social networks with MoteRunner motesCheck background information above, as well as a similar project: 12.lsi.8.This open-ended project aims to evaluate social networks in the real world, and will use WSN motes to do so. The evaluation system will be composed by three modules:
The goal of the system is to evaluate how often people are co-located, and to try to infer from that information whether they interact (e.g. by analysing how close they are, how long they are co-located, what are the characteristics of the place where they meet, etc.) A successful project will have a prototype implementation of the three modules above and will perform a number of small scale experiments with 10-20 users. Requirements: MoteRunner (EDI) Suitable for: CS, CSEmb, MEng Related work: 12.lsi.10 - Acoustic localisation using MoteRunnerCheck background information above.This project will investigate the possibility of localising a source of sound by comparing sensor readings in three different sensor nodes. It should use sound intensity and time difference of arrival to estimate the approximate position of the source of sound. Once detected, the position should be sent to a sink node connected to a PC, which then displays the estimated position of the source using a graphical user interface. The main tasks will be:
Requirements: MoteRunner (EDI) Suitable for: CS, CSEmb, MEng Related work: 12.lsi.11 - Altruistic and egoistic behaviour in wireless sensor networksCheck background information above.This project will investigate protocols that will enable altruistic and egoistic behaviours in MoteRunner motes. The idea is to allow motes to take up more workload (i.e. sensor readings, data processing and aggregation) when their energy levels are higher, but refuse and forward workload when they feel that their energy levels are too low or decreasing too rapidly. A successful project will show different ways of implementing such behaviours and evaluate the impact of having such behaviours in a deployment of 20-30 motes in terms of network lifetime and work turnaround. The actual transfer of assemblies from mote to mote and their dynamic reprogramming does not have to be part of this project, and the focus should be on the actual control of responsibilities (which should be done in a distributed way). The evaluation could be done by using dummy tasks that will (or will not) reduce the energy levels of nodes when they execute them. Requirements: very good programming and debugging skills (Java preferred), ability to understand and use a simulation environment, very good background in networks (NDS / NWC). Desired: embedded systems design (EDI) Suitable for: MEng, MSc NC Related work: 12.lsi.X - Student-defined projectI will consider intelligent student-defined projects in the areas of concurrent and distributed embedded systems. I am interested in hardware, software or system-level approaches to the design and optimization of such systems. Send me an email with a few keywords and ideas and I'll send you some feedback. |
