Skip to Main Content
Typically, cognitive radio systems either sense the channel just before transmission or perform this task periodically in order to remain aware about the operational environment. However, a channel sensed as `free' can become busy during the transmission of the cognitive system resulting in harmful collisions and unnecessary interruptions in the secondary user data transmission. As a solution, predictive based approaches has been proposed and has shown promising results in simulated environments. However, modeling real-time, dynamic, coexisting environments demand investigation with real-time demonstrators. This paper investigates industrial coexisting environments and illustrates the prediction model selection and its parameter estimation criteria. Based on the investigation a real-time testbed is implemented using a CC2500 TRX and MSP430 μC based platform.