Skip to main content

About this book

This book constitutes the refereed proceedings of the Second International EAI Conference on Emerging Technologies for Developing Countries, AFRICATEK 2018, held in Cotonou, Benin, in May 2018. The 12 revised full papers and 4 short papers were selected from 27 submissions. The papers are organized thematically in tracks, starting with ITS and security, applications and IT services, gaming and user experience.

Table of Contents


ITS and Security


Internet of Things-Based Framework for Public Transportation Fleet Management in Non-smart City

The notable increase in location-based applications especially in smart cities realm is driven by the emergence of miniaturized, cheaper and readily available location-based internet of things’ devices. The backbone of the internet of things is a well-orchestrated electronic infrastructure, telecommunication and information technology. Such a backbone is the precursor for the success of internet of things applications that have mushroomed in the public transportation sectors of the developed world. The developing countries such as South Africa have not kept pace with the development of these electronic infrastructures. Implementation of smart city concepts such as intelligent public transportation system in these countries therefore requires novel approaches. As one of the solutions to this, we present an internet of things framework that enables the integration of multiple cost-effective internet of things technologies through which public transport-related information can be obtained in cost-effective and robust ways. The framework was designed and evaluated using a system prototype for the Free State province (South Africa) public transport system case.
Muthoni Masinde, Ahmed Shoman, Mohamed Hassan Mostafa

A Middleware for Cyber Physical Systems in an Internet of Things Environment: Case of for Mobile Asset Tracking

The upsurge in Cyber Physical Systems (CPSs) has made researchers conclude that these systems have the potential of rivalling the contribution of the Internet. Driving this wave is the emergence of miniaturized, cheaper and readily available location-based hardware devices. One of the main applications of CPSs is mobile asset tracking system whose roles are to monitor movements of a mobile asset and to track the object’s current position. Localization accuracy of these systems is one of the key performance indicators. This is usually maximised through the introduction of extra hardware devices. The drawbacks with this approach include restriction of the system’s application only to one domain, introduction of extra cost to the overall system and introduction of a single point of failure. Conversely, the Internet of Things (IoT) paradigm facilitates coalescing of diverse technologies through which locus data can be extracted in cost-effective and robust way. The challenge is the lack of a dependable and responsive middleware that is capable of handling and servicing such devices. We present a solution to this problem; a middleware designed around In-lining approach that acts as an insulator for hiding the internal workings of the system by providing homogenous and abstract environment to the higher layers. The evaluation of laptop tracking and monitoring system prototype was carried out through implementation of a middleware that integrates diverse IoT components in a university environment.
Muthoni Masinde, Admire Mhlaba

Signal Processing, Control and Coordination in an Intelligent Connected Vehicle

In this paper, we present the functionalities of an intelligent connected vehicle. It is equipped with various sensors and connected objects that enable communication between the driver and its environment. This system provides assistance towards safe and green driving. The driving assistance may be directed towards the driver (semi-autonomous vehicle) or completely towards the vehicle (self-driving, autonomous vehicle). The assistance is based on the driving context which is the fusion of parameters representing the context of the driver, the vehicle and the environment. This cyber-physical vehicle has three main components: the embedded system, the networking and real-time system and the intelligent system. The architecture for data transfer within the connected vehicle is implemented through publish-subscribed infrastructure in which services are transferred and controlled in an orderly manner. These functionalities are tested both in the laboratory and on the road with satisfactory results. This is the fruit of labor of a consortium composed of five industrial and two academic partners.
Manolo Dulva Hina, Sebastien Dourlens, Assia Soukane, Amar Ramdane-Cherif

Applications and IT Services


Embedding a Digital Wallet to Pay-with-a-Selfie, from Functional Requirements to Prototype

The Pay-with-a-Group-Selfie (PGS) project, funded by the Melinda & Bill Gates Foundation, has developed a micro-payment system that supports everyday small transactions by extending the reach of, rather than substituting, existing payment frameworks. In an effort to embed a digital wallet to the PGS, we analysed the system architecture that will be needed and the requirements drive us to opting for blockchain based architecture. We have presented the applicability of a blockchain as platform in a previous paper. The current paper is presenting the functional requirements, the platforms needed for the development as well as the prototypes of the major interfaces.
Perpetus Jacques Houngbo, Joel T. Hounsou, Ernesto Damiani, Rasool Asal, Stelvio Cimato, Fulvio Frati, Chan Yeob Yeun

A Predictive Model for Automatic Generation Control in Smart Grids Using Artificial Neural Networks

This paper presents a predictive model that estimates the load for an Automatic Generation Control (AGC) system. We start by laying the foundation for our system by discussing the AGC, and the benefits of embedding it in a smart power grid. The AGC as a system is discussed with a keen focus on the mathematical relationship between the load on the system and the frequency deviation. Our predictive model is a deep neural network trained on a multivariate time series dataset for energy consumption collected over 47 months. The results show that it is possible to predict to a high accuracy, the total load on the power system within the next minute. The goal of the predictive model is predicated upon the notion that the ability to forecast the future load on the system results in the ability to estimate the frequency deviation as well, and thus giving the AGC the ability to forecast risks such as a system overload.
Chika Yinka-Banjo, Ogban-Asuquo Ugot

Exploring Users’ Continuance Intention Towards Mobile SNS: A Mobile Value Perspective

The functionalities of most Social Networking Sites allow users to enjoy practical benefits like maintaining important social and business relationships, communicating with others, and getting feedback on important shared information. However, the place of SNSs as a source of entertainment and enjoyment is also well-documented. The purpose of the paper is to identify the factors that predict continuance use of social networking sites from the perspective of mobile value. Data was collected from 452 students in three leading universities in Ghana and analyzed with Partial Least Square-Structural Equation Modeling. Results from the study revealed that both hedonic value and utilitarian value were significant predictors of continuance intention. Satisfaction was also found to be a significant predictor of continuance intention. In all, the model accounted for 55.6% of the variance in continuance intention. The study also provides important contributions to the literature, by demonstrating the significance of both utilitarian and hedonic value in leading to satisfaction with the usage of mobile SNS services. The implications and limitations of the current study are discussed and directions for future research proposed.
Aseda Mensah, Kwame Simpe Ofori, George Oppong Appiagye Ampong, John Agyekum Addae, Affoue Nadia Kouakou, John Tumaku

Smartphone Cyber Security Awareness in Developing Countries: A Case of Thailand

Cyber security awareness among smartphone users is becoming one of the main challenges of cyber security in both developed and developing countries. This paper focuses on Thailand, a developing country that is ranked among the riskiest countries in the world with regards to cybercrime. Through a survey exploring the knowledge and practices of Thai student smartphone owners, as the young population is the largest user group, we seek to estimate the level of their cyber security awareness about the most common risks. The findings reveal that they are most susceptible to identity theft or data compromise, while they were on the whole found to have a higher level of security awareness than students in other countries. We argue for Thailand’s digital economy to be sustainable, ICT4D projects need to extend their focus to this population of smartphone users to increase security awareness.
Feren Calderwood, Iskra Popova

Development of an Artificial Neural Network Model for Predicting Surface Water Level: Case of Modder River Catchment Area

Water is vital for life; however, water is a scarce natural resource that is under serious threat of depletion. South Africa and indeed the Free State is a water-scarce region, and facing growing challenges of delivering fresh and adequate water to the people. In order to effectively manage surface water, monitoring and predictions tools are required to inform decision makers on a real-time basis. Artificial Neural Networks (ANNs) have proven that they can be used to develop such prediction models and tools. This research makes use of experimentation, prototyping and case study to develop, identify and evaluate the ANN with best surface water level prediction capabilities. What ANN’s techniques and algorithms are the most suitable for predicting surface water levels given parameters such as water levels, precipitation, air temperature, wind speed, wind direction? How accurately will the ANNs developed predict surface water levels of the Modder River catchment area?
Jandre Janse van Vuuren, Muthoni Masinde, Nicolaas Luwes

UmobiTalk: Ubiquitous Mobile Speech Based Translator for Sesotho Language

The need to conserve the under-resourced languages is becoming more urgent as some of them are becoming extinct; natural language processing can be used to redress this. Currently, most initiatives around language processing technologies are focusing on western languages such as English and French, yet resources for such languages are already available. Sesotho language is one of the under-resourced Bantu languages; it is mostly spoken in Free State province of South Africa and in Lesotho. Like other parts of South Africa, Free State has experienced a high number of non-Sesotho speaking migrants from neighboring provinces and countries. Such people are faced with serious language barrier problems especially in the informal settlements where everyone tends to speak only Sesotho. As a solution to this, we developed a parallel corpus that has English as a source and Sesotho as a target language and packaged it in UmobiTalk - Ubiquitous mobile speech based learning translator. UmobiTalk is a mobile-based tool for learning Sesotho for English speakers. The development of this tool was based on the combination of automatic speech recognition, machine translation and speech synthesis. This application will be used as an analysis tool for testing accuracy and speed of the corpus. We present the development, testing and evaluation of UmobiTalk in this paper.
John Nyetanyane, Muthoni Masinde

Short Paper Session


Embedding a Digital Wallet to Pay-with-a-Selfie, Defining the System Architecture as Blockchain Based

The Pay-with-a-Group-Selfie (PGS) project, funded by the Melinda & Bill Gates Foundation, has developed a micro-payment system that supports everyday small transactions by extending the reach of, rather than substituting, existing payment frameworks. PGS is designed to work with devices with limited computational power and when connectivity is patchy or not always available. Once the concept of PGS has been accepted as demonstrated by the experimentation, we move to integrating elements and tools intended to ease federation or incorporation of the large spectrum of stakeholders. Embedding a digital wallet is one step in that vision. We analysed the system architecture that will be needed and the requirements drive us to opting for blockchain based architecture. We are then presenting the applicability of a blockchain as platform.
Perpetus Jacques Houngbo, Joel T. Hounsou, Ernesto Damiani, Rasool Asal, Stelvio Cimato, Fulvio Frati, Chan Yeob Yeun

Practical Method for Evaluating the Performance of a Biometric Algorithm

This paper presents a modality-independent method of evaluating the performance of an algorithm in biometrics. The operation mode is about developing a JAVA application that offers the user a graphical representation of the evaluation results. This application is interacting with a MySQL database containing the extracted signatures as well as the matching values of the modalities present in the evaluated biometric system. The evaluation system is used to generate the Genuine Matching and Impostor Matching score distribution curves, the False Match Rate and False Non Match Rate curves and the ROC curve. 1000 lines of code were used to develop the application. The method proposed is original and practical. Thus, an application of this method has been made in the case of a contactless fingerprint modality. We plan to improve the developed method by adding the representation of 4 main operating points (EER, WER, Fixed FMR, Fixed FNMR).
Tahirou Djara, Abdou-Aziz Sobabe, Macaire Bienvenu Agbomahena, Antoine Vianou

Software Defined Networking (SDN) for Universal Access

Ensuring Universal Access/Universal service to the populations of developing countries is up to now a big problem which can be explained by the fact that the telecommunication operators estimate certain areas unprofitable. The Universal access to the Technologies of Information and Communication being a non-discriminatory right for any citizen wherever he lives, different approaches are implemented to guarantee it. One of these approaches is to recourse on cheap equipment associated with innovating technologies. The aim of this article is to be able to study to what extent the Software Defined Networking could be a viable solution for the localities interested in Universal access. To reach this goal, we have been lead to study a typical Voice over IP traditional architecture and the architecture tooled with the Software Defined Networking technology.
When implemented the Software Defined Networking technology is supposed to guarantee a good quality of service. In our contribution we have set up a Voice over IP environment with Asterisk server as equipment of the network core, and affordable equipment such as the WIFI access points in the element entitled, ≪Collecting subscribers≫. The Quality of Service being our preoccupation, the comparison of all our results shows that the architectures with Software Defined Networking offer a better quality of services.
Adama Nantoume, Benjamin Kone, Ahmed Dooguy Kora, Boudal Niang

Gaming and User Experience


Awale Game: Application Programming Interface and Augmented Reality Interface

Awale game is one of the famous board games from Africa with many variants and is now played worldwide in various forms. In this paper, we propose an open-source Application Programming Interface (API) for developers to allow an easy implementation of the various variants of Awale as well as artificial intelligence based players. The API is available online at https://​github.​com/​Machine-Intelligence-For-You/​Awale. Based on this API, we propose a PC Awale game, a mobile Awale game, and an Augmented Reality Game. The Awale API, PC game, and mobile game are implemented in the programming language Java while the game in Augmented Reality is realized with the C# programming language, Unity 3D game engine and the Vuforia Augmented Reality SDK. The various tests carried out show that the API and the different games are totally functional. This API was also used for the first edition of MAIC, an Artificial Intelligence contest https://​mify-ai.​com/​maic2017/​.
Marie-Parisius Dorian Houessou, Vinasetan Ratheil Houndji, Eugene C. Ezin, Manhougbé Probus A. F. Kiki, Harold Silvere Kiossou, Jean-Baptiste Maureen Sossou, Faizath Jedida Zoumarou Walis

Vector Space Model of Text Classification Based on Inertia Contribution of Document

The use of textual data has increased exponentially in recent years due to the networking infrastructure such as Facebook, Twitter, Wikipedia, Blogs, and so one. Analysis of this massive textual data can help to automatically categorize and label new content. Before classification process, term weighting scheme is the crucial step for representing the documents in a way suitable for classification algorithms. In this paper, we are conducting a survey on the term weighting schemes and we propose an efficient term weighting scheme that provide a better classification accuracy than those obtening with the famous TF-IDF, the recent IF-IGM and the others term weighting schemes in the literature.
Demba Kandé, Fodé Camara, Reine Marie Marone, Samba Ndiaye

Mood and Personality Influence on Emotion

Animated conversational agents (ACA) are intended to be used to establish a relationship and communication between human and machine. In order to make credible interactions, one of important elements is to equip them with a personality that will be able to vary their behavior and social attitudes in the image of the human. This article presents the concepts and tools of personification of the machine for simulating mood and personality influence in other recognition systems of emotion.
Tahirou Djara, Abdoul Matine Ousmane, Antoine Vianou

Two Parallelized Filter Methods for Feature Selection Based on Spark

The goal of feature selection is to reduce computation time, improve prediction performance, build simpler and more comprehensive models and allow a better understanding of the data in machine learning or data mining problems. But the major problem nowadays is that the size of datasets grows larger and larger, both vertically and horizontally. That constitutes challenges to the feature selection, as there is an increasing need for scalable and yet efficient feature selection methods. As an answer to those problems, we present here two effective parallel algorithms developed on Apache Spark, a unified analytics engine for big data processing. One of them is a parallelized algorithm based on the famous feature selection method called mRMR. In the second algorithm we propose a totally novel metric to select the more relevant and less redundant features. To show the superiority of that algorithm we have created its centralized version that we have called CNFS_Spark.
Experimental results demonstrate that our algorithms achieve a great performance improvement in scaling well and take less time than classical feature selection methods.
Reine Marie Ndéla Marone, Fodé Camara, Samba Ndiaye, Demba Kande


Additional information

Premium Partner

    Image Credits