Background: Currently, spectrum sharing policies specify inflexible spectrum access controls
such as exclusion zones and maximum transmit powers based upon statistical characterizations of
all potential operating environments. The rules for preventing interference are influenced by low probability
events and are set at conservative operating levels. Consequently, spectrum sharing potential in
the majority of operating conditions is limited by a small percentage of relatively rare conditions, which
results in inefficient spectrum usage.
Objective: This paper adopts a probabilistic view of dynamic spectrum sharing systems. Specifically,
establishing in situ probabilistic reasoning reasoning within the context of a specific spectrum access
environment enables sharing systems to adapt to local conditions and uncertainties. Regulators and
spectrum users can establish spectrum access rules defining acceptable levels spectrum access risks.
Method: The paper develops a probabilistic reasoning model using Bayesian Networks to encode the
phenomenology of the communications channels and systems, and also represent the cause--‐and--‐
effect nature of dynamic spectrum sharing. This approach allows systems to infer permitted actions
(e.g., maximum transmit power) based on incomplete knowledge and established interference risk
Results: Simulation results demonstrate that situation--‐specific probabilistic reasoning combined with
risk--‐constrained spectrum access potentially enables greater spectrum sharing potential. Sharing systems
can increase transmit powers for greater channel capacity or increase network density over existing
spectrum sharing approaches. Situational uncertainty can also be reduced as sharing systems observe the
operating environment, leading to further performance gains.
Conclusions: The results of this paper demonstrate significant potential for probabilistic reasoning and
risk constrained spectrum sharing for increasing spectrum access.