Abstract: Security is of critical importance for real-time applications running on pervasive computing environments.Security-critical task model,security overhead model and security risk model are built for real-time periodic tasks,and a security risk-aware adaptive scheduling algorithm is proposed.According to inherent properties of real-time periodic tasks,the scheduling problem is transformed to a multi-stage decision-making procedure with the purpose of minimizing security risk.Based on approximate dynamic programming policy,the proposed algorithm is designed with security performance guaranteed and low-complexity mechanisms.Experimental results show that the proposed algorithm can significantly reduce security risk of applications,satisfy the security requirements,and adapt itself to dynamic pervasive computing environments.