The state-of-the-art network anomaly detection systems are often unable to correctly verify source of anomalies; they have occurred in situ or in transit. To address this issue, this study proposes an agent-enabled anomaly detection system in which mobile agents perform in situ verification of sensor nodes to verify source of anomalies. Moreover, mobile agent cannot be freely transmitted over the network, as transmission is several hundred times expensive operation as compare to processing. We optimize agent transmission through 2-sigma method. The formal behavior analysis of 2-sigma method is performed using Petri net theory to analyze its correctness. In future work, we aim to exploit cross-layer (XL) interaction to efficiently transmit mobile agent and perform quantitative analysis through experimental study.