Research On Technologies Of Data Protection Based On Trustworthy Edge Computing

The Artificial Intelligence of Things(AIoT)systems surrounding us generate,transmit,and process enormous amounts of data in a fully connected and high-bandwidth environment that integrates edge computing and faster network technologies.Through the terminal-edge-cloud-user model,it realizes the one-party sharing of data based on centralized services.However,centralized services can not realize the mode of multi-party consensus including users,services,and regulators.With the development of blockchain technology,the requirements of multi-party sharing of data,verification,and calculation are put forward to achieve the trusted sharing of data in the Internet of Things(IoT).Nevertheless,IoT has some weaknesses,including heterogeneous devices and limited resources,as well as poor environmental security,resulting in serious privacy disclosures during sharing processes.For the resource-constrained IoT,key technologies such as lightweight algorithms,data source assurance,on-chain and off-chain interactions are urgently needed to build active-defense clusters with distributed subnets and intelligent terminals at the edge.This new paradigm can protect the authenticity of data sources,and protect the privacy and property of data during sharing processes.In sight of this,this dissertation details research on key technologies of data protection based on trustworthy edge computing.Specifically,addressing the requirements of trustworthiness,authenticity,and multi-party consensus of data in the IoT,this dissertation focuses on the research of blockchain lightweight algorithm,blockchain data source authenticity guarantee,and data properties protection technology.According to the four stages of swarm consensus—namely,electing leaders,sorting blocks,verifying consensuses,and uploading chains—a lightweight blockchain model is proposed for deployment in the IoT with limited resources,including hierar-chical trustworthiness consensus(HRAFT),timestamp multi-ledger block(TMLB),and reinforcement proof of work(RPoW)algorithms.The lightweight blockchain achieves anonymous ledgers for data storage with block aggregation throughout the terminal-edge-cloud structure.Moreover,this blockchain model integrates an anti-collision broadcast voting consensus method(AC-UDE),a multi-factor verifiable consensus method(OTMP-P2L),and an autonomous network data security situation awareness method(ANSA)to seek a trade-off between computational complexity and information security.The main innovations of this dissertation can be summarized as follows.(1)Lightweight and regenesis blockchain with smart collaborative and progressive evolution based on edge computingAiming at computation and resource constraints in the heterogeneous IoT,a lightweight blockchain with smart collaborative and progressive evolution(SCOPE)based on edge computing is proposed to overcome obstacles in efficiency,preservation,and recycling of local data.By intelligent edge collaboration,the scheme asynchronously caches and submits data layer by layer throughout the three-tier structure of terminal,edge,and cloud.In particular,it submits blocks compactly,classifies a multi-ledger for a parallel uploading chain with fixed timestamp blocks,and dynamically programs different security levels for scalable services.As a result,the scheme builds a localized private chain with higher transaction speed.Aiming at the dilemma between the trustworthiness of centralization and the efficiency of decentralization of the blockchain,a reliable algorithm based on HRAFT is proposed for the daily election and sorting of leaders.The algorithm conducts a two-stage Stackelberg model of a dynamic game with complete information.It can effectively achieve the two-stage alternating Nash equilibrium of the leader and workers.Consequently,the nodes service each other and restrict each other for the accounting rights and voting rights in edge networks.Meanwhile,according to the metrics of the hierarchical activity of history,such as electing,caching,sorting,verifying,and uploading,the SCOPE can trim the high resource consumption,such as those for the proof of work(POW)and signature in the serial accounting,so as to achieve a schedulable progressive RPoW.The experiments demonstrate the peak-and-valley filling effects of the dynamic scheduling algorithm,and obtain the optimized scheduling parameters by simulation.(2)Anti-collision merge-confirmation scheme based on Bluetooth Low Energy(BLE)for voting consensusTo solve the serious signal collision of edge nodes caused by the ultra-density deployment of various short-distance communication devices in the IoT,a consensus scheme for anti-collision broadcast voting(AC-UDE)based on merge confirmation is proposed.This scheme adopts short-distance,high-speed wireless broadcast with an edge node based on a BLE improvement to control the broadcast process.In particular,the improvement protocol is a single broadcast response protocol with optimizing parameters to achieve dense voting and confirmation processing.Experimental results show that the scheme can reduce the redundant packages and channel occupation time by half to avoid more than 90%of the collisions.Therefore,it can double the capacity and decrease power consumption,and improves the efficiency of the broadcast voting consensus of the SCOPE.(3)Verifiable consensus protocol with one-time association multitasking proofs based on zero-knowledge proofAs the devices in the IoT have many issues such as weak passwords,easy forgeability,and easy disclosure of privacy,a multi-factor verifiable consensus protocol with one-time association multitasking proofs(OTMP-P2L)based on zero-knowledge proof is proposed.Specifically,the scheme builds a distributed localized security center for independent control in the edge that provides privacy-preserving,hash-identity authentication,and authorized access by zero-knowledge proof.The multitasking protocol adopts more zero-knowledge proofs than the traditional token and can more dynamically adapt to the different security levels for the heterogeneous IoT.Meanwhile,it flexibly combines multiple factors to realize a millisecond-level verification for concise P2P consensus by prefabricating tokens in batches.The experimental results show that this decentralization protocol based on the offline verification of localized ledger reduces the transmissions of sensitive information in the IoT.Moreover,the protocol reduces the manual inputs and interactions of communication,and improves the verification performance of the SCOPE.(4)Fusion network data security detection based on edge computingTo improve the detection rate of the leaks of edge data and privacy,a scheme of fusion network data security detection based on edge computing is proposed.This scheme adopts edge computing to isolate network accesses and designs a solution called autonomous network data security situation awareness(ANSA)to intelligently detect local network traffic.The ANSA constructs a real-time detector,which securely stores and integrates multi-source and multi-modal data by an edge gateway.By automatically triggering the audits of the SCOPE such as open rules,contractual sharing,and traceability clues,the households can independently be aware of their data leaks.Experimental results show that the scheme can associate multi-source data for iterative reinforcement learning of expert rules and machine-learning models.Moreover,it can find and eliminate the conflict set of multi-target recognition for fusion detection by constructing hierarchical counterfactual rules that improve the detection rate of disclosure events of data and privacy.In summary,the proposed algorithms enjoy the challenges of adapting to the heterogeneous IoT scenarios and break through the key technologies such as the lightweight blockchain model,efficient consensus,and active defense.As a result,the smarter IoT can manage data rights trustworthily,and detect data risks autonomously.Based on this solid cornerstone,the innovative schemes will promote trusted data sharing by multi-copy blocks throughout hierarchical storage in the terminal-edge-cloud structure.