Thursday, 24 August 2017

Post Graduate Researcher Details - Reginald Ankrah

My name is Reginald Ankrah. I am a doctoral researcher at The Robert Gordon University, Aberdeen. My research area concerns the use of computational techniques to optimise and solve real world problems in industries.
Telecommunications networks are vast and complex structures that need to undergo continual evolution to adapt to change. Telecoms technologies continually undergo rapid change. In parallel, user demand on networks is subject to ongoing growth periodically punctuated by the emergence of disruptive new products and services.
The market is highly competitive and so providers need to adapt promptly to demand. However, changes to the network involve considerable capital investment. This leads to complex investment decisions bearing significant financial, technical and reputational risk. In particular, to maximize return, new telecommunications networks must be designed to be robust enough to meet expected demands over their lifetime.
Computational intelligence offers a range of algorithms and techniques that are applicable to complex, multifactorial decision making. Probabilistic models derived from analysis of datasets can confer predictive insight into how decision outcomes are affected by interactions between various input factors.  They can also support Monte Carlo simulation of future decisions allowing what-if analysis, optimized decision-making and statistical confidence intervals for complex decision scenarios. Clustering approaches are used to identify similarities in datasets, for example, customer profiling. Optimisation algorithms like genetic algorithms, particle swarm optimisation and ant colonies optimisation can be used to optimize use and placement of resources when there are complex and conflicting decision factors involved. For example, such approaches would be useful in determining the placement of equipment and network structure.
My research aims at investigating the application of novel and advanced computational techniques to create specialised planning and risk-modelling algorithms focused on key aspects of network design.