2nd IEEE UK&I YP Postgrad STEM Research Symposium - Schedule

Schedule table

Time	Session	Venue
08:50-09:20	Registration Sign in and collect your symposium welcome pack. Tea and coffee will be available upon your arrival in LG.04 Foyer, Liberty Building	Liberty Building LG.06 Foyer
09:20-09:35	Welcome Address A short welcome address – Prof Robert Kelsall & Dr Matt Ritchie	Liberty Building LG.06
09:35-10:15	Symposium Keynote Address 1 Prof Mike Hinchey, IEEE UK & Ireland Section Chair “Building resilient space exploration missions”	Liberty Building LG.06
10:15-10:30	Tea & Coffee Break An opportunity to network over the coffee break.	Liberty Building LG.06 Foyer
10:30-11:50	Oral Presentation Group 1 See who's presenting in Group 1	Liberty Building LG.06
12:00-13:30	Lunch and Poster session An opportunity to network and browse the symposium poster submissions over lunch.	School of Electronic and Electrical Engineering - Foyer
13:30-14:10	Symposium Keynote Address 2 Prof Peter Jimack “Efficient numerical algorithms for the solution of fluid dynamics problems.”	School of Electronic & Electrical Engineering - Rhodes Lecture Theatre
14:10-14:40	Workshop 1 Prof Mohsen Razavi “How to give a technical presentation.”	School of Electronic & Electrical Engineering - Rhodes Lecture Theatre
14:40-15:10	Workshop 2 Prof. Christoph Wälti “How to write research fellowship applications.”	School of Electronic & Electrical Engineering - Rhodes Lecture Theatre
15:10-15:30	Tea & Coffee Break & Poster Session An opportunity to network and browse the symposium poster submissions over the coffee break.	School of Electronic & Electrical Engineering - Foyer
15.30-16.10	Symposium Keynote 3 Eric Hawthorn, Executive Chairman, Radio Design Ltd “Cellular Radio – Business & Technology.”	School of Electronic & Electrical Engineering- Rhodes Lecture Theatre
16:10-17:00	Oral Presentation Group 2 See who's presenting in Group 2	School of Electronic & Electrical Engineering- Rhodes Lecture Theatre
17:00-17:40	Panel Discussion Prof Ian Hunter, Chair. Discussion on increasing impact of academic activities on industry.	School of Electronic & Electrical Engineering - Rhodes Lecture Theatre
17:40-18:00	Closing Remarks, Awards Ceremony Awards will be presented for the best oral and poster presentations. Both awards are sponsored by IEEE Communications Society (ComSoc) Chapter UK and Ireland.	School of Electronic & Electrical Engineering - Rhodes Lecture Theatre
18:40	Symposium Networking Dinner A chance to network with symposium attendees and speakers over a 3-course dinner and welcome drink. (interested participant will pay a charge to the restaurant).	Leeds Red Hot buffet

Speakers should be prepared to make an 8-minute presentation with 2 minutes allocated for questions (10 minutes total).

---

Oral Presentation Group 1

Chaired by Dr Mustafa Bakr

Time	Speaker
10:30 - 10:40	Oluseyi Temitope Okeowo
10:40 - 10:50	Valentina Casadei
10:50 - 11:00	Francois S. Hallac
11:00 - 11:10	Carla Taylor-Rutterford
11:10 - 11:20	Nishal Chandarana
11:20 - 11:30	Ali K Abed
11:30 - 11:40	Ahmed Fakhri
11:40 - 11:50	Conor Dorrian

Oluseyi Temitope Okeowo - University of Salford

Autonomous navigation of a mobile robot using a Histogram filter The model presented in this work illustrates the performance of a probabilistic filtering algorithm for the autonomous navigation of a mobile robot in a structured environment. We selected a grid-based application of Bayes’ filter, the histogram filter, for the study. A mobile robot needs to understand its present location, where the goal is, and how it can reach the goal to achieve autonomous navigation. The problem of localization comes first, followed by path planning and navigation. Our model uses a histogram filter to localize the robot in a closed environment, and later perform path planning through dynamic programming to navigate to a target location. The histogram filter discretizes the robot’s world into a grid-world setting and assigns a probability value to each cell. We preferred the odometry motion model to the velocity motion model in calculating the robot’s motion over time. Though both models suffer from drift, odometry model is generally more accurate. We selected a Pioneer 3DX mobile robot as the test bed owing to its robustness in performing autonomous navigation tasks. The results of the experiment presented in this paper using MobileSim and the real robot validates the accuracy of the proposed model.

Valentina Casadei - University of Liverpool

EEG signal processing with a smart model-based tting: a comparison with traditional methods Electroencephalography is a safe recording of brain activity, used in clinical and non-clinical purposes. Extracting meaningful information from EEG poses significant challenges for digital processing due to non-stationary behaviour and a very small Signal-to-Noise Ratio. Such challenges have become particularly important with wearable EEGs, used without clinical supervision. They are more sensitive to noises and are generally less accurate than clinical devices. A quantitative evaluation of the measurement uncertainty is essential to understand the acquisition reliability. A good signal processing must achieve two objectives: 1) work without changing the original shape and information of the raw signals avoiding the risk of a wrong interpretation; 2) continuously and automatically assess the uncertainty of any quantity estimated from the signals. Some benefits are a decreased risk of wrong classification and an improvement for feature selection and extraction, especially for a dynamic system. This can be achieved by using available ‘a priori’ knowledge about signals (frequency band and dynamics of amplitude and phase modulations) to implement a model-based fitting process able to follow the time evolution of specific brain waves, estimating its uncertainty and reliability simultaneously. This model works as automatic smart filtering, with better performance than common signal processing methods in the time domain (band-pass filtering) or frequency domain (Fourier Transform).

Francois S. Hallac - University of Leeds

The breakage of elongated particles for agitated drying conditions in the pharmaceutical industry Crystal breakage is an issue of great concern to the pharmaceutical industry. Conservation of the desired Particle Size Distribution (PSD) throughout downstream processing is extremely important, as PSD changes are known to affect properties such as bulk density, solubility and flowability. Active pharmaceutical ingredients are in majority organic and their crystal shape of high aspect ratio; their main breakage mechanism is fragmentation by bending. To elucidate the fracture phenomenon of elongated particles in agitated drying, the bending stress of individual crystals needs to be determined within a bed of particles. In this study, a shear cell is built in Discrete Element Method and mimics the stress experienced by particles in dryers using moving parallel plates and periodic boundaries. Elongated rigid particles are modelled with clumped spheres and experience stress due to the shear application in the box. The bending stress of individual particles is calculated during the simulation and a bending stress distribution is obtained for the given stress condition (normal and shear). The bending stress distribution is combined with the experimental breakage strength distribution of Beta-Glutamic Acid crystals, allowing the estimation of the extent of breakage within the particle bed during the shearing phase. The extent of breakage is found to increase exponentially with the hydrostatic pressure for agitated drying conditions in the pharmaceutical industry.

Carla Taylor-Rutterford - University of Lincoln

Tele-Sex: Opportunities, Challenges, and Implications There is growing interest in utilising IoT and haptic technology for intimate communication. Specifically, where two devices are networked with sensors and effectors which enable intimate interaction and communication. This has flourished since the expiration of the patent for the ``method and device for interactive virtual control of sexual aids using digital computer networks’’ (U.S. Patent No. 6,368,268.), otherwise known as the teledildonics patent. However, this application is rife with privacy, security, and ethical challenges, which are confounded by a general lack of research in this specific domain. This review provides a map of the current research engagement and identifies and illustrates some of the potential risks (with some historical context), and an exploration of the future implications of this technology. The range of possible benefits and drawbacks in regard to this technology are varied, highlighting the need for transdisciplinary engagement. We conclude with a number of insights, and a call to arms for further research.

Nishal Chandarana - University of Sheffield

Decarbonisation of heat: a heat pump load shifting model The UK has now committed to reducing greenhouse gas emissions to net zero by 2050. To do so requires significant progress to be made in the heating sector, which accounts for a quarter of overall emissions. Heat demands are highly variable, reaching huge peaks during and hourly ramp rates during winter. The natural gas network can currently cope with this variability, but a range of solutions will be required to transition to a low carbon system. Electrification of heating, using heat pumps, is one attractive option due to the significant progress in decarbonising electricity generation. Concerns remain, however, over impacts that widespread uptake could have on peak electricity demands, particularly on the distribution network level. In this study the operation of a heat pump and thermal storage system has been optimised to minimise costs using a variable electricity tariff that discourages peak electricity usage. The potential peak demand reduction has been quantified and a comparison has been made between the building material and thermal storage unit in shifting usage. Results indicate that building thermal mass and therefore heat pump scheduling plays a significant role in shifting electrical load to off peak times.

Ali K Abed - University of Bradford

Development of improved second-generation of the Automated Solar Activity Prediction system (ASAP) In recent years, there has been a growing interest in real-time processing of solar data, especially for space weather applications. Solar flare prediction has become a forefront topic because extreme solar eruptions could affect our daily life activities. Therefore, the main goal of this paper is to develop a new prediction algorithm based on the Automated Solar Activity Prediction system (ASAP) system. The proposed algorithm updates the ASAP system by extending the training process and optimizing the learning rules to create better optimization for the performance. Two neural networks are used in the proposed system. The first neural network is used to predict whether a certain sunspot class at a certain time is likely to produce a significant flare or not. The second neural network is used to predict the type of this flare, X or M-class. Some measurement criteria are applied to determine the extent of system performance and all results are provided in this paper. The results exhibit that the proposed algorithm outperforms the old ASAP system.

Ahmed Fakhri - Teesside University

Techno-Economic Analysis of Li-ion Batteries in the Capacity Market with Different Degradation Models Increased deployment of intermittent renewable energy plants raises concerns about energy security. Capacity markets (CM) have been implemented to secure energy generation and provide energy affordability for customers. This paper contributes to answering the question of whether batteries can provide cost effective back up services in this market. It improves on earlier research in the field by considering battery degradation and capacity market de-rating factors. To provide a comprehensive techno-economic analysis, three lithium ion battery degradation models with different complexity were utilised. The results demonstrated that degradation cost can significantly impact the potential profit from each battery and the battery with the 1h de-rating factor shows the highest revenue within the current CM regulations. The empirical degradation model is simple but overestimate the capacity loss especially in the first cycles. Keeping the temperature at 5ºC and at low SoC (20%) offers the highest profit. In contrast, the semi-empirical model shows that the degradation cost is maximum at 5ºC and minimal at 25ºC with SoC (20%). The physics model offers a deeper understanding for the complex degradation mechanisms inside the battery and it shows that minimum degradation is at 5ºC in storing condition. While CM rules require batteries to remain ready to respond at system stress events, the results suggest that large losses are anticipated if batteries are kept at 100% state of charge.

Conor Dorrian - Letterkenny Institute of Technology (LYIT)

Can event-driven micro-services be ameliorated through performance feedback from performance data? When an event-driven micro-service is created there is often less emphasis into performance testing or monitoring it as testing isn’t straightforward and proves to be very time consuming. This may result in errors and difficulties later in the lifecycle of the event-driven microservice as it scales. Performance testing should be a necessity when developing a microservice, but unfortunately it does not often happen unless a serious complication arises further in the event-driven lifecycle, this will take up more time, energy and cost than doing it at the start of the deployment lifecycle. Each service in a microservice may have its own dependencies, some direct, others transitive and each service could connect to its one or more databases such as Mongodb and MySQL. Establishing a valid record of these dependencies for each service is difficult to maintain, requiring a dedicated team to monitor each service inside the microservice. Whether it’s database errors, network latency, caching issues, or service unavailability, event-driven microservices should be able to handle a reasonable level of faults. Performance testing highlights defects and performance issues in the event-driven microservice early and ultimately will help curb these issues before going into a production environment. The aim of this paper is to inform development teams and DevOps engineers about managing their performance testing within microservices. Using specific tools to test different aspects of performance testing such as Load testing and resilience testing will ultimately create a more fault tolerant microservices from the evidence collected through the means of performance testing to create a tangible and adaptable framework.

---

Oral Presentation Group 2

Chaired by Dr Thaddeus Eze

Time	Speaker
16:10 - 16:20	Salem Mansour
16:20 - 16:30	Sheena Worthington
16:30 - 16:40	Muhammad Saleheen Aftab
16:40 - 16:50	Simon Obute
16:50 - 17:00	Menwa Alshammeri

Salem Mansour - University of Sheffield

A Meta-Analysis for Evaluating the Efficacy of Different BCI Designs for Upper Limb Stroke Rehabilitation Recently brain-computer interface (BCI) has attracted increasing attention as a potential tool for post-stroke upper limb rehabilitation. Nevertheless, different BCI designs, used in clinical trials, stated different clinical outcomes. The mechanism and efficiency of BCI for stroke rehabilitation generally remains unclear, and it could be hard to conclude the effective BCI paradigms for upper limb stroke rehabilitation. In this Line, we present for the first time a comprehensive meta-analysis to evaluate the short-term and long-term efficacy of BCI for the hemiparetic stroke rehabilitation, as well as the efficacy of different aspects of the BCI design according to the type of performed mental tasks, applied neural classification features, and feedback mechanism given to patients. Out of 587 studies, 11 studies matching the inclusion criteria were found. The upper limb Fugl-Meyer Assessment (FMA-UE) scores from the selected randomized control trials (RCTs) were pooled, and the overall effect sizes were estimated. In summary, our results support the use of band power feature, neuromuscular electrical stimulation and the intention of movement in the future BCI design for the upper limb post-stroke rehabilitation. However, this conclusion should be taken carefully since our sample size was relatively small and the heterogeneity between the selected RCTs studies.

Sheena Worthington - University of Chester

New Chemical Processing and Recycling Technologies for Auto-catalysts The purpose of this research project is to develop a method to recycle ceramic fibre. The ceramic fibre to be investigated is from catalytic converters. The current methodology involves the characterisation of the feed ceramic material, this will establish a suitable composition to investigate further, such as; recovering of aluminium, silicon, magnesium, calcium compounds from the large quantity of ceramic material. And lastly, to validate the promising evidence gathered from the Bayer process principles. An examination of several samples of the ceramic insulation fibre revealed that reduced particle size of milled ceramic fibre is more powerful than the particle size of unmilled ceramic fibre, the milled particles are more exposed and has more chances for individual coverage and also has a higher possibility for the solvent and particulate matter to respond. The effects of metal dissolution do not have a significant impact with the results of the samples of ceramic insulation fibre, even no string device was integrated into the design of the apparatus for both acid and alkali digestion experiment. The particle size was determined by the use of SEM analytical technique. In addition to this, the autoclaving time is also varied, by contrasting two sets of time; 24 hours versus 72 hours. As no drastic change observed between two sets of time, further analysis was made with 24 hours due to it being more economical. The Carbolite Gero oven with hydrothermal autoclave was used as a conventional heating at 120°C. The sulphuric acid precipitation with 10M NaOH shows a good crystalline state. The determination of the chemical compounds’ quantity is underway. The use of the three experimental techniques such as; XRPD, ICP, EDS, suggests the presence of the Zeolites (NaAlSi2O6-H2O). Currently the proposal scheme is still in its novelty. The future success of the project depends on quantity recovered, how pure it is, and the available market for the products recovered.

Muhammad Saleheen Aftab - University of Sheffield

On handling difficult dynamics with Predictive Functional Control Predictive Functional Control (PFC) offers numerous advantages such as trivial coding, easier implementation and simpler handling without requiring sophisticated knowledge or specialised personnel. Besides systematic handling of input and output constraints as well as process dead-times makes PFC a preferred control strategy over more conventional techniques, say PID, backed with many successful industrial applications. PFC, arguably, belongs to the simplest branch of model-based predictive control (MBPC) family wherein the principal design is based on some rather simplistic assumptions, for instance, plant output behaving like first-order dynamics, and the manipulated variable being fixed within prediction horizon. Although the resulting control law may not be optimal, but it is effective enough for satisfactory closed-loop performance and that too without requiring computationally expensive algorithms. Nevertheless, these simplifications render PFC ineffective in applications when challenging dynamics, such as open-loop instability or significant under-damping, are prevalent. Obviously, the problem is with constant input, which lacks flexibility to cope with divergent or unsettled predictions and simultaneously provide smooth set-point tracking. Reported modifications usually parametrise input to overcome unwanted dynamics, but at the price of increased cost and coding complexity. A less common alternative is the use of explicit pre-stabilisation or pre-conditioning loops that may provide smooth and settled prediction behaviour. This talk presents two pre-compensation methods namely compensation via (i) pole shifting and, (ii) pole cancellation, and discusses their advantages and disadvantages with regards to set point tracking, constraints management and handling of external perturbations such as input and output disturbance, sensor noise and model uncertainties. The work is pretty much in initial stages and the presenter would highly appreciate constructive feedback in this regard.

Simon Obute - University of Leeds

RepAtt: Achieving Swarm Coordination through Chemotaxis Swarm foraging is a common test case application for multi-robot systems. In this paper we present a novel algorithm for improving coordination of a robot swarm by selectively broadcasting repulsion and attraction signals. Robots use a chemotaxis-inspired search behaviour based on the temporal gradients of these signals in order to navigate towards more advantageous areas. Hardware experiments were used to model and validate realistic, noisy sound communication. We then show through extensive simulation studies that our chemotaxis-based coordination algorithm significantly improves swarm effectiveness when performing a foraging task. A maximum reduction in foraging time of 83% was achieved in one world configuration.

Menwa Alshammeri - University of Leeds

Enhancing Topic Modeling for Quran Using Paragraph Vectors Having a good representation of short text like Quran can benefit the semantic understanding and inferring coherent topics. The Quran has been the subject of numerous studies due to its linguistic and spiritual value. Scholars have studied Quran for its topics. They have drawn out knowledge that was the base for many applications to allow search in the holy book. Many studies were devoted to topic modelling with the Quran. Latent Dirichlet Allocation was mainly adopted in these works. However, they were limited to a unigram model. Most research projects focused on the translation of the Quran in different languages instead of the original text. In my work, I use the original text of the Quran. Besides, I explore using verses of the Quran as the input for the topic modelling algorithm. Analysing a text like the Quran requires learning approaches that go beyond word level to achieve phrase or sentence level representation. So, in this work, I explore using paragraph vectors to learn vector representations of Quranic Verses. These vectors can be used as features and leveraged for the topic analysis. I then test Paragraph vectors on the task of finding related verses of the Quran. I have observed that verses that have a high probability under the same topics tend to be highly semantically related to each other. The positive results confirm that having good representations for texts can promote its understanding and locate the semantically related verses under the same topic.

---

Poster Presenters

Presenter	Title of Presentation
Kieran Bull	Systematic construction of scarred many-body dynamics in 1D lattice models
Francesco Foglino	Curriculum Learning for Cumulative Return Maximization
Omar A. Jasim	A framework of Control System Verification for Flights Safety of Unmanned Aerial Vehicles
Francois S. Hallac	The breakage of elongated particles for agitated drying conditions in the pharmaceutical industry
Nikollao Sulollari	Terahertz imaging of surface plasmons on 2D structures
Flora Biggins	Optimal scheduling of energy storage devices to create profit through arbitrage
Menwa Alshammeri	Enhancing Topic Modeling for Quran Using Paragraph Vectors
Majid Al Saadi	An Analysis of Teaching and Learning Strategies Used in Mathematics Classrooms in Jordan
Mohamad Riduwann bin Md Nawawi	Electrical Characterization of GaAsBi MWQ pin diode for Future Photovoltaic Cell Application
Syeda Fariha Hasnain	Critical Analysis and Exploration of Artificial Intelligence Algorithms and Techniques in CRN
Eleanor Nuttall	Terahertz Lasers: New Techniques for Studying the Chemistry of Climate Change
Alexia Toumpa	Relational Graph Representation Learning for Predicting Object Affordances
Christopher Wirth	"You Have Reached Your Destination": A Single Trial EEG Classification in Navigation
Pavel Israel Muniz Zavala	Table tennis ball trajectory prediction and control through deep learning and neural networks.
Sophie Middleton	Towards digital citizenship: A digital literacy curriculum to support teachers in the classroom
Hammajam Ahmed Adamu	Security and Privacy Compliance Framework for Software as a Service (SaaS) Business Applications – Retail Sector of the Nigerian Oil and Gas Industry as a Case Study
Lukasz Tyszczuk Smith	Multimodal Physiological Markers for Improving Diagnosis and Prognosis of Depression
Samantha Sargeant	Artificial Intelligence Technology for Stem Cell Therapy Manufacturing
Ahmed Azab	An Ensemble Framework with Temporal Alignment for Improving BCI Performance in Small Sample Settings
Yu Liu	Optogenetic Brain-machine Interface System for a Visual Cortex Prosthesis
Sheena Worthington	New Chemical Processing and Recycling Technologies for Auto-catalysts
Soham Gharia	AgriBlockIoT: Towards traceability in future food-supply industry
Bintang Ekananda	Energy Audit and Energy Conservation Measures Analysis using eQUEST – A Case Study at Neville Hill Train Maintenance Depot, Leeds, UK.
Henrique Gil	Serious games and empathy
Mubashra Latif	Improving technically challenging biomass fuels for heat and power production