LMR spectrum bands are used both for commercial and non-commercial purposes servicing first responder organizations such as police, fire, and ambulance services, public works organizations like utility companies, dispatched services such as taxis, or other companies with large vehicle fleets. The FCC primarily allocates LMR channels for voice communications through commercial and federal or non-federal government agencies/services. Land Mobile Radio (LMR) band that range from 138 to 941 MHz, would potentially be a good candidate for implementing Dynamic Spectrum Access.
Moving even small bands of spectrum to a dynamic access regime cannot be done manually. There is a need for an intelligent and learning based system such as artificial intelligence techniques to support real time prediction of spectrum usage. Cognitive Radios (CR) appear to be a potential solution to solve the problem of inefficient spectrum usage and spectrum scarcity among unlicensed. Perhaps the first step in establish in a CR environment is to come up with a reliable prediction mechanism that indicates those channels which are going to be suitable for transmission based on the previous usage data available.
Occupancy of three randomly selected Land Mobile Radio Band channels over 20 days
Aim: We set out to investigate the benefit of the “memory” of long short term memory (LSTM) networks in predicting spectrum occupancy in multiple time horizons in Land Mobile Radio (LMR) bands.
Background: ANNs are a popular choice for spectrum prediction. Traditionally, ARIMA models have been at the forefront of forecasting. However, recurrent ANNs have demonstrated good prediction performance.
Methodology: We train and evaluate four prediction models: a baseline which simply delays the time series, a seasonal ARIMA model, a TDNN and an LSTM. We test their performance on an hourly dataset in LMR bands collected in Ottawa, Canada between the dates of October 2016 and April 2017.
Results: We find that LSTMs provide an improvement in prediction performance compared to the other models. We also compare the computational complexity of our models. Conclusions: The LSTM networks that remember long term dependencies and designed to work with time series provide an improvement accurately predicting spectrum occupancy in LMR bands.