Research On Collaborative Privacy Computing For Mobile Sensor Dat

Mobile sensing,as a new pervasive sensing mode,has become the main part of intelligent ubiquitous sensing and computing.Technological innovations such as multi-domain deep integration,complicated application scenarios,cross-domain data collaboration and distributed computing collaboration have brought many challenges of security,accuracy and efficiency to sensing computing.Therefore,the research of effective mobile sensing collaborative privacy computing technology has important theoretical research significance and application value.Among them,how to deconstruct the privacy requirements in mobile sensing data aggregation computing,how to deeply explore the incentive game problem of multi-party sensing collaboration,how to optimize privacy aggregation strategy to achieve task-based selection aggregation,and how to design more effective cryptographic protocol for multi-scenario collaborative privacy computing applications,these are the key issues that need to be solved urgently.Therefore,this paper focuses on three aspects of privacy incentive and game learning,privacy aggregation and security sharing,and cryptographic collaborative computing,propose a sensing incentive mechanism for dynamic reputation optimization,an aggregation strategy with optional equivalence(interval)judgment,a cryptographic sparse sensing truth calculation protocol and a cross-platform sensing collaborative security recommendation algorithm,which formes a series of key technologies and innovations achievements,enriches the theoretical system and methods of mobile sensing collaborative privacy computing,and can realize the privacy security control of the whole process of mobile sensing data aggregation computing.The main research results of this paper are as follows.(1)Aiming at the persistent incentive problem of privacy cooperative computing in the process of mobile sensing data aggregation,a dynamic incentive strategy optimization algorithm based on game learning is constructed.Firstly,combined with the interactive characteristics of sensing participation in multi-stage strategies,a dynamic two-stage reverse auction incentive model is defined,an alliance payment contract based on reputation mechanism is designed.Furthermore,the group sensing game model is constructed from the perspective of cooperative incentive,the utility function of mixed income and quantifying privacy cost is designed,the Nash equilibrium characteristic of maximizing expected fairness is analyzed,and the optimal privacy aggregation strategy is adopted to achieve the balance between user privacy cost and cooperative computing utility.Secondly,combined with the dynamic evolution theory,the game recursive learning rule is constructed,and the privacy aggregation strategy optimization algorithm based on reputation improvement is proposed to achieve the progressive equilibrium state of dynamic decision-making in group game.Finally,the simulation experiment and numerical analysis of strategy and utility evolution show that the proposed model and method can solve the contradiction between individual rationality and alliance cooperation,effectively realize the privacy incentive of cooperative participants,and achieve the goal of group aggregation fairness strategy optimization.(2)Focusing on the problem of data acquisition redundancy in massive sensing data aggregation,a novel fine-grained cryptographic sensing is designed.Firstly,the aggregation strategy cryptographic validation method based on Boolean decision function is designed,constructed logic expression of aggregation strategy through the linear secret sharing.Aiming at multiple attribute equivalence aggregate,polymerization algorithm though cleverly introducing virtual master key is designed to realized selective aggregation without decryption,it can improved the safety of the aggregation scheme.Furthermore,by effectively combining KP-ABE and BGN homomorphic encryption algorithms,the cryptographic equivalent attributes selectable aggregation scheme is proposed.Secondly,according to the more complex requirements of interval attribute aggregation,a range matrix is constructed by combining the idea of binary tree encryption,defined a publicly computable function embedding key generation algorithm.A cryptographic aggregation determination scheme suitable for interval attributes is proposed.Finally,through provable security theory,the proposed scheme is proved to have semantic security under selective plaintext attack,and has significant advantages of fine-grained aggregation operation compared with existing schemes,it can achieve the cryptographic selectivity and cryptographic computability under different computational task requirements.(3)Aiming at collaborative privacy computing application scenario of sparse sensing truth reconstruction,cryptographic sensing truth computing protocol is proposed.Firstly,a truth reconstruction problem is defined formally for the sparsity of sensing data,an auxiliary computing server is introduced to construct cryptographic sensed computing architecture based on semi-honest model,which can perform complex function computation of private data according to the sensing computing task.Furthermore,based on the homomorphic encryption algorithm,designed a variety of basic secure computing protocols,such as secure distance calculation and secure KNN query,which can support the protocol design of truth reconstruction cryptographic cooperative privacy computing.Secondly,a polynomial fitting method is proposed to solve the cryptographic calculation problem of the similarity weight function.On this basis,a truth reconstruction privacy calculation scheme is designed to ensure the computational security and anonymity under the semi-honest model.Finally,theoretical analysis and experimental simulation are given from the aspects of protocol security,algorithm computational cost and calculation accuracy,the good calculation accuracy is shown in the application of large-scale environment-aware regional truth fitting,which can meet the security computing needs of sensing truth reconstruction.(4)Aiming on the application requirements of cross-platform sensing collaboration security service recommendation,an efficient secure multi-party collaborative filtering algorithm based on hash technology is proposed.Firstly,through hamming mapping of the matrix transformation and perturbation factor designing,constructed an local sensitive hash algorithm meeting the privacy protection mechanism,proposed a cross-platform similarity assessment model with sensing preferences,in order to realize the collaborative recommendation clustering security index mechanism,it can improve privacy calculation efficiency through the hash collision neighbor sampling data dimension reduction method.Secondly,in order to meet the security and accuracy requirements of cross-platform collaborative service recommendation,based on the secure multi-party cooperative privacy computing framework of arithmetic secret sharing mechanism,a two-party hidden vector matrix factorization secure collaborative filtering algorithm is designed based on the neighborhood index.Finally,the performance evaluation and comparative analysis of a large number of simulation experiments show that the proposed scheme has significant advantages in both computational accuracy and performance,which can ensure the security of cross-platform collaborative computing and efficiently realize the sensing collaborative accurate service recommendation.