HomePublicationsTechnical Report

Self-Organized Consensus as a Key to drive Pervasive Internet-of-Things
Ref: CISTER-TR-191203       Publication Date: 13, Nov, 2019

Abstract:
Nowadays, mobile networks are suffering from the capacity crunch in meeting users demands. The situation will get even worse in the future due to the deployment of a massive number of networked objects (e.g., self-driving vehicles, smartphones, wireless devices, and embedded objects), as predicted by network experts forecasting. All these objects have sensing capabilities and produce various types of data, frequently timedependent, and the operational expenses for managing all these networked objects are anticipated to be very high. Therefore, future networks must provide fast access to small up-to-date data produced by a vast set of objects, which may have intermittent availability. This is typically called the Internet-of-Things (IoT). The future networks for supporting the IoT must provide lower latency, higher energy efficiency, intelligent security, and increased reliability whichever users’ mobility and availability patterns, beyond the current state-of-the-art network connectivity. Will it be possible to develop future networks with all these properties while reducing operational expenses? Our thesis is that the answer is positive if the future networks exhibit properties of self-organization, also called as autonomic functions (self-optimization, self-recovery, intelligent security, and self-configuration). Self-organized networked (SON) systems are capable of performing complex tasks via coordination and cooperation, which cannot be achieved by individual autonomous objects. The coordination of SON systems gets their foundation from the consensus theory. Even though each system has a different behavior, it can lead to the emergence of a global intelligent behavior by following simple communication rules. The performance of consensus protocols is a vital concern for SON systems, namely convergence time/rate, robustness, and fault-tolerance. However, consensus protocols must be aware of the mutual interference among autonomic functions accordingly, aiming to support anticipation. This thesis work aims at investigating the multi-agent systems (MASs) based coordination algorithms, biologically-inspired, and quantum-inspired approaches for supporting SON systems. The MAS-based coordination algorithms are built upon the basis of the composition of multiple cooperative self-organized agents, that can solve problems that are difficult or impossible for an individual agent. Biologically-inspired approaches are based on the principles of biological evolution of nature; which is concerned with the intelligent behavior of biological species by simple and local interactions among individuals. The quantum-inspired approaches employ quantum mechanics logic to be executed on classical computers, which are exponentially faster due to a high degree of parallelism. Therefore, the proposed MAS-based, bio-inspired, and quantum-inspired consensus approaches can provide high data rate, low latency, energy efficiency, reliability, and intelligent security despite the availability and mobility patterns of heterogeneous objects. To tackle these issues, first, we propose a distributed and intelligent architecture by addressing the physical connectivity and communication protocols. Then, by relying on the proposed architecture, we define a set of autonomic functions (i.e., self-optimization, self-recovery, and intelligent security) and a common representation of associated information for achieving intelligent security, energy efficiency, and Quality-of-Service. Based on such standard description, consensus algorithms must be applied to ensure that autonomic functions cooperate while pursuing different goals. The performance of such consensus processes, and so of the mobile networks, may be improved by performing anticipation of network operations based on user’s availability and mobility patterns, as well as on the performance of individual autonomic functions.

Radha Reddy