Efficient structured policies for admission control in heterogeneous wireless networks
read more
Citations
Blocking probabilities of elastic and adaptive calls in the Erlang multirate loss model under the threshold policy
Performance metrics of a multirate resource sharing teletraffic model with finite sources under the threshold and bandwidth reservation policies
A hybrid (N/M)CHO soft/hard vertical handover technique for heterogeneous wireless networks
Structured Admission Control Policy in Heterogeneous Wireless Networks with Mesh Underlay
An Erlang multirate loss model supporting elastic traffic under the threshold policy
References
Markov Decision Processes: Discrete Stochastic Dynamic Programming
Markov Decision Processes
A First Course in Stochastic Models
Vertical handoffs in wireless overlay networks
Related Papers (5)
Optimal Dynamic Multicast Scheduling for Cache-Enabled Content-Centric Wireless Networks
Frequently Asked Questions (15)
Q2. How can the authors reduce the computation cost of evaluating the cost function in every step?
When times between decision epochs are exponentially distributed the authors can reduce the computation cost by introducing fictitious decision epochs at which no real decision has to be made.
Q3. What are the common methods used to solve MDPs?
Several methods such as Value Iteration (VI), Policy Iteration (PI) and Linear Programming (LP) methods are developed to solve general MDP problems [8].
Q4. What are the common optimality criteria in CAC?
The most common ones are minimization of a total cost (objective) function and minimization of the blocking probability given some hard constraints on dropping probabilities.
Q5. What is the method for determining the optimal admission policy?
efficient numerical methods called Structured Value Iteration (SVI) and Structured Update Value Iteration (SUVI) are proposed to determine the optimal admission policy.
Q6. How can the authors find the average cost of a given MDP?
Once the authors have the average cost the authors can use methods such as multidimensional bisection search [22] to find the parameters that minimize it.
Q7. What is the average service time for a call?
Service requests (more specifically calls in this work) arrive according to a memoryless Poisson process, and also service times are memoryless.
Q8. What is the threshold for admission to overlay?
If there is an î for which ∆iVk (̂i, j0) ≤ CLR and ∆iVk (̂i+1, j0) > CLR, then î is the threshold for admission to overlay when there are j0 calls in underlay.
Q9. What is the ratio of BR area to RR area?
As the network capacity increases, the ratio of BR area, in which full optimization is performed, relative to the area of AR/RR regions, in which a default action is evaluated, decreases.
Q10. What is the idea of the SVI algorithm?
In SVI, the authors assign a default action to every such point based on the region it belongs to, and then in every round of iteration the cost function for that point is updated according to that default action.
Q11. What is the idea of the proposed SVI algorithm?
under the operation of a numerical algorithm similar to SVI, in every iteration the changes in cost or decisions can only happen within the border region.
Q12. What is the overall result of the optimal control scheme?
The overall result is that the optimal control scheme does not allow for a linear change in system rejection costs to be reflected severely in the average cost.
Q13. what is the optimal cost function for a system model?
The authors also assume the following boundary conditionsVk(Cc + 1, j) = ∞ and Vk(−1, j) = 00 ≤ j ≤ Cw Vk(i, Cw + 1) = ∞ and Vk(−1, j) = 00 ≤ i ≤ Cc. (4)The authors show that the optimal policy to minimize the average cost for the system model given in Section II is a 2D thresholdbased policy.
Q14. What is the cost of rejecting a call request?
The authors can formally define MINOBJ asMINOBJ : min gπ = ∑L k=1 C (k) R λkP (k) B(1)where C(k)R is the cost of rejecting a call request of class k, λk is the arrival rate of class k calls, P (k) B is the blocking (dropping) probability for that class and L is the total number of call classes.
Q15. What is the fictitious call event type?
The authors add a fictitious call event type of 0 which corresponds to call departures with a fictitious decision of a = 0 to be taken at departure events.