无线移动互联网：原理、技术与应用崔勇CH2APAssociation课件教学.pptVIP

* PPT研究院 POWERPOINT ACADEMY * * * * * * * This is a service area map for a three story building with 60 demand clusters. Each cluster represents not a single user. In fact, it represents group of traffic requests. This is a signal level map for a three story building with 14 candidate APs. AP is placed in a way that no coverage gaps exist in the service are. On the first floor, we equally divide the service are and put APs. The signals from first floor APs travels to second floor, but signal gets weaken because of walls and distance. Therefore, coverage hole happens on the 2nd floor. We put APs on those holes. We repeat the procedure, which we did on the first floor. If we put too many APs, the cost of installation will be high, If we put too less APs, the adequate coverage will not be guranteed. We formulate the problem of AP selection and traffic allocation by minimizing the congestion of the most heavily loaded APs. Our objective is that we are minimizing the maximum congestion of APs. Subject to, First, each demand cluster should be assigned to only one AP in (2). Next in (3), it defines the congestion factor of APs. Summation of this is the total traffic request from demand clusters within threshold ranges to a APj. Congestion of AP j is the total traffic request asked to AP j divided by Bj, which is bandwidth capacity of AP j. If C j = 0, then that means AP j is not selected by any demand cluster. By minimizing bottle neck APs, we can get better throughput for the whole network. The proof is by contradiction, using Corollary 3. The intuition behind it comes from Equ. (1) in Slide 34. The minimal normalized bandwidth allocation is maximized (max-min fairness) when the maximal load on the APs is minimized (min-max load). The proof is by contradiction, using Corollary 3. The intuition behind it comes from Equ. (1) in Slide 34. The minimal normalized bandwidth allocation is maximized (max-min fairness) when the maximal load on the APs is minimized (min-