vii Secondly, the feasibility of using neural network forecasters to detect spoofed sampled values was investigated. We showed that although such forecasters have high spoofeddata-detection accuracy, they are prone to the accumulation of forecasting error. In this research, we proposed an algorithm to detect the accumulation of the forecasting error based on lightweight statistical indicators. The effectiveness of the proposed algorithms was experimentally verified on the Smart Grid testbed at FIU. The test results showed that the proposed techniques have a minimal detection latency, in the range of microseconds. Also, in this research we developed a network-in-the-loop co-simulation platform that seamlessly integrates the components of the smart grid together, especially since they are governed by different regulations and owned by different entities. Power system simulation software, microcontrollers, and a real communication infrastructure were combined together to provide a cohesive smart grid platform. A data-centric communication scheme was selected to provide an interoperability layer between multi-vendor devices, software packages, and to bridge different protocols together. viii