Research Of Non-Local Sensing Methods Based On Collaborative Computation Of Localized Information

In nature,the changes of the world always have signs to follow under the basic mathematical-physical laws.Will this happen in the information world?If so,how to efficiently track the occurence and movements of these changes?Could our human beings get rid of traditional costly data collection methods,and be able to get a glimpse of the whole autumn picture via a leaf?Be more specific,as the rapid development of mobile Internet,Internet of Things(IoT),and cloud computing,distributed sensing,computing and storage have been widely applied in various fields,with decentralization as the core design concept.How-ever,there is a fundamental conflict between the information bottleneck limited by the local view in decentralization/localization and the global information needs for effi-ciency optimization.These global information such as user distribution,virtual machine interconnections,and routing topologies often play a critical role in optimization and decision making.However,it is difficult to collect and use these dynamically changing information in distributed systems.With the wide application of distributed systems,such as Internet of Things(IoT),Blockchain,etc.,the local information bottleneck in decentralization/localization are increasingly becoming the bottleneck for solving many problems in these fields,where the difficulty lies in the following two aspects.·Timeliness.It is usually time consuming to collect the global information to the base station.Therefore,the collected information at the BS often no longer re-flects the most recent state of the system.Therefore,it is difficult to use these stale information to achieve expected effect.·Overhead.As mentioned earlier,the collection of global information relies on the aggregation of local information traversing through individual nodes,which is costly in communication and storage.Particularly,when requiring continuous acquisition of global information,it is almost impossible for the system to afford the communication and storage cost.This paper addresses the fundamental contradiction between the naturally existing local information bottlenecks of distributed systems and the need for continuous ac-quisition to global information,and based on the recent frontier advances in computer science,number theory,physics and other related fields,reveals the potential orderli-ness and causal correlations of the topological changes of connected systems between number theory and space-time domain,mapping these information correlation and inter-action from physical space/information space to high-dimensional computation space,and proves that the evolutionary trajectory in the phase space caused by any arbitrary changes is causal and traceable.It get rid of the constraints of traditional information ac-quisition,and establishes a new distributed sensing theory based on local information to infer the current and historical global information:Non-locality Sensing Theory,which first time makes the interesting effect generally practical in the information world:"get a glimpse of the whole autumn picture via a leaf ",We have provided rigorous proof and shown the correctness and efficiency with detailed mathematical analysis.Meanwhile,this paper explores a series of related research areas such as distributed sensing,net-work security,distributed consensus of blockchain and brain-like storage,and finally proposes a series of sensing and storage methods in related fields.Specifically,the main research content of this article is divided into the following parts:1.At first,this paper focuses on the global topology and identity information of the connected network,and provides a localized computation paradigm to map those global information with bytesize chaotic code at each node.This local-ized paradigm computationally builds a transformation of identity and topology from time and space domain to a new computation domain.Such a transfor-mation maintains four properties:transitivity,global convergence,determinacy,and causality.Based on these properties,this paper proposes an algorithm for inferring global information based on byte size scale of local information to re-cover and continuously track global topology and ID information.To our best knowledge,it is the first time that enables a network to track and infer the global information without relying on the traversal transmission of metadata in the net-work.2.In security area,this thesis also provides new detection methods against clone attack and sybil attack based on the non-locality sensing theory,and achieves the deterministic detection.Compared with traditional methods,the proposed detection methods outperforms in detection accuracy,communication overhead,storage overhead,and privacy protection.3.The paper further extends the non-locality sensing theory and proposes a new dis-tributed consensus mechanism.It generalizes the global tracking and inference mechanism from topology information to more general information,for track-ing arbitrary "transaction data",and no longer rely on multi-stage submission and leader election,correspondingly provides a new mathematical foundation blockchain.Finally,This paper implements and builts a blockchain transaction ledger system with proposed mechanism,and initially verified the deployment feasibility of this blockchain prototype system.4.Furthermore,note that the non-locality sensing theory provides a novel informa-tion encoding way with structured association mapping the global information between the space and time domain and the computation domain.Such an en-coding way could further be applied for data storage,with nearly constant stor-age space complexity but infinite storage capacity.The proposed design reveals a new feasible way for efficient storage and information interaction based on sim-ulated impulse neurons,and first time provides a viable mathematical theoretical support for Nobel Prize winner Burt Sackman's brain-like memory conjecture.