Titulo Estágio
TelcoPredict: Machine Learning Techniques for Cell Usage Prediction
Áreas de especialidade
Engenharia de Software
Sistemas Inteligentes
Local do Estágio
Coimbra, Taveiro
Enquadramento
Telecommunications is becoming a commodity and with that Telco companies have to deal with the constant dropping of the average return per user (or customer). This fact makes the Telco companies to be more efficient, trying to cut down costs. One of the areas that Telco companies invest a lot of resources is in the mobile cell network, which makes its planning a crucial task, ensuring that the network antenas are not to much or not too low for the usage that the nodes have. Critical Software has developed a platform for managing the mobile network structure, which enables some form of planning of the network deployment resources. But having a way to forecast the cell usage according to the location, time of the day, month day, and other variables is crucial for a better improvement of the cell deployment. The employment of forecasting techniques from the machine learning and Artificial Intelligence area is a promising way yo tackle this problem.
Objetivo
The main goal of this internship is to develop a set of forecasting algorithms capable to predict the number of customers that are going to use a specific cell of a mobile network. The available information resides in the Telco company that provides historical usage information so that Machine Learning algorithms can be used to create forecasting models. The main idea is to defined a solution based on machine learning and/or other Artificial Intelligence techniques able to solve this forecasting problem. This goal can be subdivided in:
- Defining the Scope of the Cell Usage problem and Understanding the Available Data and System
- Creating the Technical Specification
- Development of the Solution inside Critical Software Solution
- Testing and Benchmarking the Solution against real data
Data understanding and preprocessing is crucial for the application of Machine Learning or other computational learning methods. Another main challenge of this internship is to develop a performant solution, since the volume of information is huge and needs to be processed very fast.
Plano de Trabalhos - Semestre 1
Fases do estágio: descrição, resultados, calendarização
The internship has the following stages:
- Defining the Scope of the Forecasting Problem [result: requirement list, September and October]
- Understanding the Available Data and System [result: Data and System description, September and October]
- Reading and Writing the State of the Art [result: state of the art, September to December]
- Creating the Technical Specification [result: technical specification, December to February]
- Writing the internship proposal [result: internship proposal, January and February]
Plano de Trabalhos - Semestre 2
Fases do estágio: descrição, resultados, calendarização
The second semester comprises the following stages:
- Development [result: first prototype, February to May]
- Testing and Benchmarking [result: second prototype, June]
- Writing the internship report [result: internship report, June and July]
Condições
O Orientador Académico e o Orientador Industrial são responsáveis por acompanhar o Estagiário garantindo que este tem as condições necessárias para a execução do estágio, incluindo acesso a instalações e materiais ne-cessários para o efeito. A avaliação do Estágio é da responsabilidade da Instituição de Ensino Superior, sendo o Orientador Industrial responsável por prestar informações requeridas por esta para esse efeito.
A bolsa de estágio oferecida é de 450 euros.
Orientador
Bruno Saraiva
bj-saraiva@criticalsoftware.com 📩