Belief functions and uncertainty management in networks and telecommunication
- 1.2k Downloads
In the last few years, Dempster–Shafer theory also known as Theory of Belief Functions (TBF) or Evidence theory has received growing attention in many fields of applications such as finance, technology, biomedicine, etc. This theory may be seen as a generalization framework of different instances such as probability, fuzzy sets, and possibility theories. Using Dempster–Shafer belief functions to express available information allows considering two kinds of uncertainty: aleatory uncertainty due to the variability of the variable of interest in the population and epistemic uncertainty due to a lack of knowledge on the state of the variable.
Different sources of uncertainty and imprecision may arise in network and telecommunication domains. Such imperfection may be due to imprecision of many aspects regarding the environment: signal, data link, network, etc. For example, it may be due to communication links that might be unreliable, either due to operational tolerance levels or environmental factors. As detailed in the survey paper proposed par Mustapha Reda Senouci, Abdelhamid Mellouk, Mohamed Abdelkrim Senouci, and Latifa Oukhellou in this special issue, the Theory of Belief Functions has proved to be particularly useful to represent and reason with partial information in a wide range of applications, including signal processing, coding, supervision, localization, resource provisioning, etc. In such case, the belief function theory provides a flexible framework for handling and mining imprecision and uncertainty as well as combining different disparate evidence about uncertain events. Indeed, this theory allows modeling different concepts such as imprecision, ambiguity, and ignorance. Also, a variety of combination operators is available in the fusion process.
This special issue of Annals of Telecommunications is intended to provide the recent advances on the use of the Theory of Belief Functions and machine learning approaches in telecommunication and network technologies. It focused on how belief functions and machine learning have affected different aspects (protocols, algorithms, paradigm, energy, signal coding, etc.) for a large family of applications (healthcare, medical, underwater, vehicular, robotic, etc.) using network technologies (sensor networks, MANET, VANET, etc.).
This special issue starts with papers dedicated to the general problem of uncertainty management in telecommunication and networks. They cover several topics as varied as traffic identification, routing, and synchronization/decoding. Frederic Launay and Patrick Coirault address the problem of synchronization and decoding of quasi-chaotic signals. They propose a new approach based on belief propagation algorithm. A second-order Markov model is used to synchronize a quasi-chaotic sequence. Two algorithms are considered, namely, the Viterbi algorithm and the backward-forward algorithm. Based on a combined use of multifractal analysis of wavelet energy spectrum and neural network classifiers, Hongtao Shi, Gang Liang, and Hai Wang propose a novel traffic identification approach which does not require any payload information. Different application traffic identifications are achieved by performing classification over the wavelet energy spectrum coefficients that were inferred from the original traffic.
The problem of human activity recognition in the framework of belief functions is then investigated by Faouzi Sebbak, Farid Benhammadi, Abdelghani Chibani, Yacine Amirat, and Aicha Mokhtari. In this paper, the authors present a framework based on Dempster–Shafer Theory to tackle the problem of human activity recognition in smart home environments. The proposed methodology allows converting and combining the raw data captured using a wireless sensor network into high-level activity knowledge. It allows also optimizing decision-making in the presence of conflicting activities.
Considering vehicular ad hoc networks, Mira Bou Farah, David Mercier, Eric Lefevre, and Francois Delmotte propose two models based on belief functions for exchanging and managing uncertain events on the road. The main purpose of this application is to provide the most reliable information for the driver from multiple uncertain messages received about events on the road. The proposed approach allows to manage imperfect information about events on the road in vehicular ad hoc networks. The experiments are carried out through various simulations. An implementation with Android smartphones using a Bluetooth technology to exchange the messages is also introduced in the paper.
The remaining papers address the problem of routing. A novel energy-aware clustering algorithm for the Optimized Link State Routing (OLSR) is proposed by Ahmed Loutfi, Mohammed Elkoutbi, Jalel Ben Othman, and Abdellatif Kobbane. The authors present a clustering approach that elects a reduced and reasonable number of cluster heads that have a high residual energy. Their solution can prolong the lifetime of the entire network and enhance the routing process. The paper proposed by Nadeem Javaid, Ayesha Bibi, Akmal Javaid, Zahoor Ali Khan, Kamran Latif, and Mohammad Ishfaq investigates the quality routing metrics in wireless multi-hop networks. The authors consider three existing quality link metrics, namely, Expected Transmission Count (ETX), Minimum Delay (MD), and Minimum Loss (ML), and they propose a new one, called Inverse Expected Transmission Count (InvETX), with the aim to enhance the conventional Optimized Link State Routing (OLSR) protocol. Simulations are carried out to compare all these quality metric results for conventional and enhanced OLSR.
The last paper, proposed by Abdullah-Al-Wadud, Md. Abdul Hamid, and Ilyoung Chong, investigates the particular problem of fault-tolerant structural health monitoring protocol using wireless sensor networks in the framework of Dempster–Shafer theory. By discounting the unreliable observer’s measurements, the authors show significant improvements in terms of detection accuracy compared to bayesian approaches.
We thank all anonymous reviewers for their hard work, time, and support that greatly helped us select the best papers for this special issue. We also thank all authors who submitted their papers for the consideration for this issue. We express our gratitude to the staff of Annals of Telecommunication for their support and kind encouragements throughout the preparation of this work. Finally, we hope you will enjoy reading this selection of papers as we did.