A method and apparatus for spectrum access by secondary users in a dynamic spectrum access (DSA) system are provided. In a DSA system, a primary user (PU) has priority access to the channel. A secondary user (SU) can transmit opportunistically when the channel is idle, i.e., when the PU is not transmitting. However, the SU should vacate the channel when PU re-appears. Thus, the dynamic access of an SU is a prediction problem. Given that a channel has been idle for duration t, should an SU be allowed to transmit, and if so, for how long, such that the probability of the SU interfering with a PU is below a given threshold. We present a stochastic method based on survival analysis to solve the prediction problem. Unlike other DSA schemes in the literature, our method does not assume any particular PU occupancy distribution. The proposed method builds an estimate of the cumulative hazard function in the learning phase which is later used to predict the duration of SU transmission. Our method uses this non-parametric estimate of the cumulative hll7Jll'd function to predict the SU transmission duration, which makes the method very simple to implement and thus, attractive for practical systems.
There is a need for government and industry to share spectrum due to limited frequency band. This invention provides for a dynamic spectrum access (DSA) in a tiered sharing system whose goal is to increase spectrum utilization. It is an apparatus (hardware + firmware) that implements a method for granting DSA to a secondary user (SU) when the primary user (PU) is idle. It gives a statistical guarantee on the probability of interference and predicts the remaining idle time available to the SU given only the duration of the idle period at the time of the SU’s request. Additionally, this invention provides a statistic based on previous observed idle periods known as the hazard function. The PU has priority access to a given frequency band. The SU can transmit during unoccupied (i.e., idle) periods on an opportunistic basis. The SU must vacate when the PU needs the band. DSA becomes a matter of prediction. It is based on the following scheme (see figure below): Given that the channel has been unoccupied by the PU for duration t and a request from an SU arrives to transmit for a duration τ, what is the probability that the SU will be able to complete the transmission before the PU re-appears on the channel? Our approach is novel in that it uses the hazard function from survival analysis – well known in analysis of biological organisms and failure in mechanical systems – to answer this question. Every unused idle period is a lost (billing) opportunity. Carriers pay lots of money to license frequency bands (e.g., AT&T $18.2B, Verizon $10.4B, Dish Network $10B and T-Mobile $1.8). This invention is a game changer.
This invention is a statistical guarantee on the probability of interference that is easy to interpret.