On the Performance of Hidden Markov Model Spectrum Opportunity Forecast on Limited Observed Channel Activity

The increasing demands for wireless channels to accommodate the surge of internet of things devices and the associated services exacerbated the need for flexible channel allocation strategies. Opportunistic spectrum sharing is expected to provide a more reasonable use of the limited radio frequencies available by allowing the coexistence of licensed users and unlicensed users in the same frequency. This arrangement is called opportunistic channel allocation, where unlicensed users explore the channel when the licensed user is not transmitting. The challenge in opportunistic spectrum allocation is to find transmission opportunities. Accurate opportunity detection mechanisms to avoid interference and improve spectrum usage are highly desirable. Hidden Markov Model training and predicting procedures are proposed in this work to balance the number of training sequences to limit the influence of outliers and provide opportunity forecast even when the training process is executed over a limited number of observed sequences. Our findings show that higher accuracy can be obtained even when the HMM is trained with a reduced number of transmission sequences. The results show that, compared to similar works, the proposed alternatives reduce collision rates while improving the overall number of seized transmission opportunities. The proposed HMM training procedures are able to identify over 90% of channel opportunities with PU load ranging from 20% to 80% of the channel capacity. Also, the collision rates, that is, when both PU and SU would be transmitting concurrently on the channel, was less than 10% for PU load in 30-90% of the channel capacity. Furthermore, the proposed HMM training procedures reduced the collision rate by 45.1% and improved the number of seized opportunities by 4.9%.