Research On Topology Change Aware And Maintenance Mechanism Based On Crowd-Sensing In Distributed Network

The fast development of Internet and widely-deployed IoT devices have triggered explosive growth of network applications.Motivated by the observation that there is an increasingly aggravated conflict between decentralized networking with self-driving paradigms/applications and expanding demands for consistent network sensing,maintenance,and management,this research aims to provide new localized computation paradigms attempting to· 1.replace global communication/storage with a purely localized communication paradigm at constant computation and communication complexity for network maintenance;· 2.build causality relationship in a new dimension(chaotic code dimension)between any pair of nodes that are originally irrelevant in the physical space(4-dimensional space,i.e.,length,width,depth,and time);· 3.achieve global perspective of the entire network(non-locality sensing)through one-hop local knowledge,via local computation with causality inference done in a new dimension.This research first presents a purely localized backbone renovating algorithm(LBR)to support network maintenance.Specifically,the proposed algorithm renovates a broken backbone in the network with ultra low communication and computation overhead,while keeping the backbone size within a constant factor from that of the minimum CDS with a guaranteed connectivity of the network.The main contributions are as below:· 1.The proposed algorithm has the ability to renovate the backbone in a purely localized manner,and the communication range is limited in local area,whose communication and computation overhead are both constant;· 2.Unless the network is no longer connected,the LBR can always keep the connectivity of backbone connected;· 3.The proposed algorithm is able to deal with arbitrary number of’ node failures/additions in the network in a purely localized way;· 4.Compared with existing schemes,the experiment results show that the proposed algorithm can maintain the renovated backbone being connected with a relatively small backbone size.Secondly,this research proposes a new topology sensing method via a localized computation paradigm.Specifically,the proposed research builds a topologically causal relevance at a new dimension between any pair of nodes that are physically(in 4-Dimension space,i.e.,length,width,depth,and time)irrelevant.The network(system)built upon the proposed design has four interesting properties:transitivity,convergence,determinacy,and causality.Once there is a topology change in a local area,there may result in a global change(butterfly effect)in a chaotic code dimension.Based on the four properties and the knowledge of the initial network state,any node may exactly perceive that change in the network.The proposed research provides a purely localized computation paradigm that restricts computation/communication/storage within onehop.A node is only responsible for sending/receiving/computing the information for itself,which means no message relay is no longer necessary for sensing commnunication any other nodes in the network.The main contributions are as below:· 1.The proposed research builds a causal relevance from 4-dimension physical space to a high-dimensional space,enabling the network(system)built upon the proposed design with four interesting properties:transitivity,convergence,determinacy,and causality;· 2.Though the proposed localized paradigm restricts a node’s knowledge within one-hop,any node may achieve global perspective of the entire network(nonlocality sensing)based on the four properties and the knowledge of the initial network state;· 3.To the best of our knowledge,it is the first time that a network is enabled with non-locality sensing ability at local knowledge without any remote data collection assistance.Besides,the proposed approach could provide continuous monitoring,in which the promptness and the communication cost could be significantly different from traditional periodical/event-driven approaches.This research further considers the threats of clone attack in Internet of Things(loT),an emerging networking paradigm in which a large number of inter-connected devices communicate with each other.Usually loT devices suffer the vulnerability of physical capture attacks.Previous approaches against clone attacks either relies on centralized designs that suffer from a high communication/storage overhead,or ground on decentralized probability based schemes sometimes with a poor detection accuracy.Besides,most of these approaches may also impose a high risk of privacy information disclosure(e.g.,location,etc.).In this research,we propose a chaotic token chain based scheme for detecting clone attacks in loT scenarios,which ensures deterministic detection of clones in a purely localized manner,while keeping location information private.Based on local neighbor characteristic,the proposed scheme computes for each device a chaotic token(code),which is updated continuously.Once the token changes,a node is able to judge whether there exists a clone node,and locates where the clone node is.The main contributions are as below:· 1.Different from traditional clone attack detection methods,the proposed detecting scheme ensures deterministic detection of clones based on the topology causal association;· 2.Compared with traditional periodical and probabilistic detection method,the proposed detection scheme can monitor clone attack continuously;· 3.The proposed scheme provides clone attack detection via a global perspective,while keeping overhead within local neighborhood at each device.