Aiming at the problem of the difficulty in privacy protection and the cost increase caused by privacy protection in the user data submission stage in Mobile Crowd Sensing (MCS), CS-MVP algorithm for joint privacy protection and CS-MAP algorithm for independent privacy protection of the attributes of user submitted data were designed based on the principle of Local Differential Privacy (LDP). Firstly, the user submitted privacy model and the task data availability model were constructed on the basis of the attribute relationships. And CS-MVP algorithm and CS-MAP algorithm were used to solve the availability maximization problem under the privacy constraint. At the same time, in the edge computing supported MCS scenarios, the three-layer architecture for MCS under privacy protection of the user submitted data was constructed. Theoretical analysis proves the optimality of the two algorithms under the data attribute joint privacy constraint and data attribute independent privacy constraint respectively. Experimental results show that under the same privacy budget and amount of data, compared with LoPub and PrivKV, the accuracy of user submitted data recovered to correct sensor data based on CS-MVP algorithm and CS-MAP algorithm is improved by 26.94%, 84.34% and 66.24%, 144.14% respectively.

To reduce the possibility of falling into the local optima during the search process of the original Particle Swarm Optimization (PSO) and avoid destroying the population diversity, a hybrid two-norm particle swarm optimization algorithm with crossover term, namely HTPSO, was proposed. Firstly, the two-norm was employed to measure the Euclidean distance between current particle and its individual history best one. Then, the Euclidean distance was incorporated into the velocity updating formula in order to affect the influence of social term on particles' velocity, and inertia weight was randomly distributed in accordance with certain rules. Based on these operations, what's more, HTPSO was simplified and the crossover operator in the Differential Evolution (DE) algorithm was incorporated into the algorithm, which enables each particle to intersect with its individual history best one under a certain probability. In order to verify the excellent performance of HTPSO, four improved PSOs were introduced, including Particle Swarm Optimization algorithm for improved weight using Sine function (SinPSO), Self-adjusted Particle Swarm Optimization algorithm (SelPSO), Mean Particle Swarm Optimization algorithm based on Adaptive inertia Weight (MAWPSO) and Simple Particle Swarm Optimization algorithm (SPSO). The optima of eight commonly used benchmark functions in different dimensions were compared, the results of five algorithms were analyzed by T-test, success rate and average iteration times. Compared with the contrast algorithms, HTPSO has strong convergence, and the particles' movements are very flexible.

In order to overcome the disadvantages of Differential Evolution (DE) algorithm such as low optimization accuracy, slow convergence and poor stability, an Adaptive Differential Evolution algorithm based on Multi-Mutation strategy (ADE-MM) was proposed. Firstly, two disturbance thresholds with learning functions were used in the selection of three mutation strategies to increase the diversity of the population and expand the search scope. Then, according to the successful parameters of the last iteration, the current parameters were adjusted adaptively to improve the search accuracy and speed. Finally, vector particle pool method and central particle method were used to generate new vector particles to further improve the search effect. Tests were performed on 8 functions for 5 comparison algorithms (Random Mutation Differential Evolution (RMDE), Cross-Population Differential Evolution algorithm based on Opposition-based Learning (OLCPDE), Adaptive Differential Evolution with Optional External Archive (JADE), Self-adaptive Differential Evolution (SaDE), Modified Differential Evolution with p-best Crossover (MDE_pBX)), and each example was independently performed 30 times. The ADE-MM algorithm achieves a complete victory in the comparison of mean and variance, 5 independent wins and 3 tie wins are achieved in the 30-dimensional case; 6 independent wins and 2 tie wins are obtained in the 50-dimensional case; in 100-dimensional case, all are won independently. At the same time, in the Wilcoxon rank sum test, winning rate and time-consuming analysis, the ADE-MM algorithm also achieves excellent performance. The results show that ADE-MM algorithm has stronger global search ability, convergence and stability than other five comparison algorithms.