SCTD: A spatiotemporal correlation truth discovery scheme for security management of data platform

The continuous development of mobile sensing devices enables them to quickly perceive a large amount of data, forming a promising mobile crowd sensing (MCS) platform for large-scale data collection. The raw data stored in the data platform is further synthesized into a variety of Internet of Things (IoT) services to data consumers through the processing of the data platform. However, due to the wide range of data sources, sensing devices may contribute corrupted data, or maliciously spread forged data, causing the data platform to make wrong decisions and damage the quality of service. Therefore, collecting high-quality data is critical for the security of the data platform and the quality of IoT applications. In this paper, a novel Spatiotemporal Correlation Truth Discovery (SCTD) scheme is proposed, which adopts historical data as verifiable evidence to identify the truth of reported data and gain the trust of workers, consequently recruiting high-trust devices to collect data. First, Unmanned Aerial Vehicles (UAVs) are sent to collect Gold Ground Truth Data (GGTD), which is used as the benchmark to verify the data truth of the minority sensing devices. Then a trust evaluation method is proposed to calculate the trust of devices. Second, the data reported by trusted devices as Silver Ground Truth Data (SGTD) is utilized to verify the trust of most devices, so the method proposed in this paper can discover the truth of massive data. Third, to reduce the cost of truth discovery, a low-cost method of data fitting is proposed to collect massive historical data of the trusted device, thereby verifying the truth of data in the same time and space. Since historical data contributes little value to IoT services, the platform can obtain a large amount of historical data by paying low rewards to the devices. Finally, we propose to select mobile sensing devices to collect truthful data in different spaces, which can effectively cover the spatiotemporal correlation data truth discovery in time and space, thereby verifying as much data submitted to the platform as possible. Based on the trust relationships constructed in this paper, a novel trust-based recruitment scheme is carried out for selecting the most trustworthy workers to participate in data-sensing tasks. The experimental results show that our solution can accurately identify the trust of more workers and verify the truth of data in a wider range while minimizing the cost of the data platform.