Improved BPA based Multi-source Nonhomogeneous Continuous Context Inconsistency Resolution Approach

In context-aware systems (CASs), because of precision variance of sensors, network delay and equipment heterogeneity, the context received from the dynamic environment usually exists the inconsistency which would result in improper services. An improved basic probability assignment (BA) based approach is presented in the paper to eliminate inconsistent contexts for multi-source non-homogenous sensors. For the multi-source nonhomogeneous continuous context (MNCC) inconsistency resolution, the BPA is innovatively defined by three factors: sensor precision, membership degree and grey relational grade (GRG). Experiment results demonstrate that the proposed improved BPA based approach is more effective in eliminating inconsistent contexts compared with the other typical context inconsistency resolution approaches and the correctness of the proposed approach is greatly improved.