Performance Evaluation

We compare the hit ratio of the proposed SCC with the random method in Fig. 6.9. In the simulation, the distribution of interests follows the Zip-f distribution [24]. The rate of content delivery is determined by the Poison distribution. The degree

Hit ratio with different simulation times of users is decided by the power law distribution

Fig. 6.9 Hit ratio with different simulation times of users is decided by the power law distribution. Here, the random method means that the priorities of replicas to cache in the content store are determined at random. We carried out experiments to test the performance under different simulation times. Here, the priorities of replicas to cache in the content store are determined by costq,k. As costq,k takes into consideration the popularity of content, features of the social community, and delay, it can be looked upon as the cost to remove cached content from the content store. From Fig. 6.9, it can be obtained that the proposed scheme can result in a better hit ratio than the random method. More details can be seen in [25].

 
Source
< Prev   CONTENTS   Source   Next >