Research On Collaborative Caching Strategies At Network Edge Based On D2D Communications

Global mobile data traffic has experienced an unprecedented increase in the past decades with rapid proliferation of smart wireless devices and the mobile internet.This growth in internet traffic is expected to continue at an even higher pace,mainly contributed by video streaming and related multimedia services.With the rise of D2 D communications,employing D2 D devices to carry out edge cache has become a key technology for unloading network downlink flow.However,due to the limited cache space of D2 D devices,it is very important to make a reasonable caching strategy.Therefore,we mainly study the collaborative caching strategies at network edge based on D2 D communications.First,a collaborative caching system at network edge based on D2 D communications is modeled,including: system model,the distribution model of requested content,the search and delivery model of content,and cache hit ratio at network edge.Based on the system model,an OSP caching strategy which can get the optimal performance of the system is introduced.Then,based on the analysis of OSP caching strategy,we propose a Random Caching on Truncated Zipf(RC-TZ),which optimizes the cache distribution by setting two cache parameters: truncated threshold and cache Zipf index.Due to the introduction of the two caching papameters,compared with OSP caching strategy,the number of parameters that need to be optimized are decreased to 2,so RC-TZ caching strategy not only reduces the complexity,but also makes the relationships among the content caching probabilities more intuitive.Next,in order to search the parameters of RC-TZ caching strategy,we introduces two jointly searching caching parameters(truncated threshold and caching Zipf index)algorithm: a steepest search method and a genetic algorithm,and study the influences of different system parameters on the values of cache parameters.At the same time,in order to further reduce the complexity,a two-step search algorithm is proposed,which searches the truncated threshold and the caching Zipf index respectively.Finally,we analyze the effects of different system parameters on system cache performances.In addition,we also compare OSP,RC-TZ caching strategy with other caching strategies.Although the complexity of RC-TZ caching strategy is decreased largely,the results show that very close performances to the performances of OSP are achieved by our proposed RC-TZ.Compared with some existing caching strategies,RC-TZ caching ategy can improve the cache performances significantly.