Spatiotemporal Prediction Based Intelligent Task Allocation for Secure Spatial Crowdsourcing in Industrial IoT

With the emergence of spatial crowdsourcing technology, an efficient task allocation is the key to ensure the sustainable development of spatial crowdsourcing. However, previous spatial crowdsourcing task allocation technologies ignore the temporal and spatial continuity between historical task data, thus reducing the efficiency of crowdsourcing task allocation. In addition, spatial crowdsourcing also suffers from the privacy leakage problem. To solve these problems, we propose a Spatiotemporal Prediction based Spatial Crowdsourcing strategy, named SPSC, using both blockchain and artificial intelligence. Specifically, considering the temporal and spatial continuity of crowdsourced task data, SPSC combines both gated recurrent unit and variational autoencoder for crowdsourcing task prediction. In addition, different Laplacian noises are added to crowdsourced task data so as to protect the privacy of crowdsourced workers during the task prediction. Moreover, by classifying crowdsourcing tasks and grouping crowdsourcing workers, SPSC reduces the risk of crowdsourcing workers colluding to steal the privacy data of crowdsourcing tasks using the blockchain technology. The experimental results show that SPSC can improve the privacy protection of spatial crowdsourcing, specifically, the more the number of categories, the higher the degree of privacy protection, and under the premise of predicting value, excellent system performance can be achieved.