چكيده به لاتين
Cognitive radio technology has been used to efficiently utilize the spectrum in wireless sensor networks. Specific features of cognitive radio (such as spectrum sensing and variation of available bandwidth due to the activity of primary users) affect the amount of data transmission in cognitive radio sensor networks (CRSNs). Duo to the constraints of CRSNs, introducing an admission control mechanism which is aware of primary users' activity can be fruitful in order to provide quality of service for multimedia applications.
In this dissertation, a threshold-based admission control mechanism is proposed which decides based on average available recourses of CRSNs. In order to decide more accurate about giving admission to cognitive radio sensors to send information based on primary user's activities, a policy-based admission control is presented and the system is modeled by a semi Markov decision process (SMDP) in this mechanism.
In this proposed policy-based admission control, some measures such as delay, power and blocking probability of data flows are considered based on requirements of CRSNs. According to these requirements, the average end-to-end delay and blocking probability constraints are considered in decisions of proposed SMDP. Furthermore, in order to enhance the network lifetime, a power-aware weighting method is considered for issued data flows of cognitive radio sensors.
In order to calculate the average end-to-end delay of the network, it is needed to calculate the average backlog of cognitive radio nodes. Therefore, the queue of cognitive radio nodes are modeled as an M/M/C with primary users based breakdown and hence, this calculated average backlog is used in the Kleinrock's independence approximation formula.
At last, the performance of proposed model for the average backlog of cognitive radio nodes is evaluated and this model is verified and the decisions of the proposed admission control mechanisms are evaluated and compared in different scenarios by different experiments which are performed by simulation experiments based on NS2 and CogNS frameworks.
