Date of Graduation
Fall 2025
Degree
Master of Science in Computer Science
Department
Computer Science
Committee Chair
Hui Liu
Abstract
Mobile Crowdsensing (MCS) is a sensing paradigm that leverages mobile devices to conduct a large-scale data collection. However, due to its openness and mobility nature, it is highly vulnerable to security issues such as injection attacks of malicious workers and fake tasks that can severely affect the platform’s normal functioning. To address this problem, the arrival of workers and task submission process is represented as a multivariate time series, and a two-stage framework is proposed. In the first step, we propose a novel transformer-based model, DozerAnomaly, that can efficiently detect anomalies in multivariate time series. We integrated a sparse attention mechanism, Dozer self-attention, with local and seasonal adaptation that captures the locality and seasonality of data patterns. It addresses the quadratic complexity of the standard self-attention mechanism by focusing only on relevant time steps, thereby reducing computational overhead. We employ the association discrepancy between prior and series associations to distinguish normal and abnormal time series patterns. The detection of malicious attackers contributes to improving the resilience and reliability of the sensing platform. In the second step, we conduct experiments by transforming task assignments into a bipartite graph. Experimental results in five real-world datasets show that our model achieves state-of-the-art performance in terms of accuracy and efficiency, with a 42% reduction in floating-point operations (FLOPs) compared to the baseline, significantly improving computational efficiency without compromising accuracy. Additionally, we perform task assignment experiments that show around 36% improvement in work and task assignment accuracy, demonstrating the effectiveness of the proposed framework in the MCS domain.
Keywords
anomaly detection, malicious workers, fake tasks, transformer, dozer attention, dozeranomaly, time series, crowdsensing
Subject Categories
Computer Sciences
Copyright
© Sanjeev Shrestha
Recommended Citation
Shrestha, Sanjeev, "Sparse Transformer for Anomaly Detection in Mobile Crowdsensing" (2025). Graduate Theses/Dissertations. 4112.
https://bearworks.missouristate.edu/theses/4112
