Research On The Cache Technology For The Near-end Services In The Internet Of Things

With the continuous development of the Internet of Things(IoT)technology,various IoT applications and services are emerging one after another,providing users with a continuous power for enjoying comfortable,efficient and convenient life,and injecting new vitality into the development of society.The IoT near-end service makes the location of information services as close as possible to end users.It can provide more efficient information services in the resource-constrained IoT environment.According to the use of resources by IoT services,it can be divided into three types:data transmission-intensive services,computing-intensive services,and comprehensive IoT services.For different types of IoT services,data caching technology,service caching technology and cache-aware computation offloading can be used to construct the efficient near-end service model.It can be seen that various ways of building near-end services are closely related to the cache technology.Cache technology can be said to be one of the cores of building the near-end service model in the IoT environment.Therefore,it has great theoretical value and practical significance to carry out in-depth research on the cache technology for the near-end services in IoT environment.This paper conducts research on different types of IoT near-end services in terms of data caching,service caching,cache replacement in computation offloading,and cache-aware offloading strategies,and proposes a series of new models and new methods:(1)For the problem of data caching in the IoT environment,we study the data caching technology under the ICN architecture.Then,we propose an approximate global optimal solution and a distributed local optimal solution based on the Lagrange heuristic algorithm.It realizes the local calculation of the data cache placement location without adding additional communication overhead.Aiming at the problem of how to ensure the freshness of periodic data,an active periodic data update mechanism is proposed based on the NDN model,which achieves the goal of minimizing network data transmission while ensuring data freshness.By constructing a simulation experiment environment based on NDNSIM,the validity of the proposed theory is verified from multiple angles such as network traffic and data freshness.Experimental results show that the locally optimal distributed cache algorithm can reduce the network traffic by 12.6%on average,and the periodic data active update mechanism can reduce the network traffic by 33.8%on average.(2)For the problem of service caching in the IoT environment,especially for the characteristics of mobility of IoT devices,we study mobile-aware service allocation and caching strategies.Firstly,we study a motion trajectory prediction model based on frequent pattern mining,and propose a service allocation algorithm based on target location in combination with the current service cache state.The algorithm achieves mobile-aware service allocation with the goal of maximizing the number of users served by the local edge server.Then,based on the short-term dynamic changes of service requests and long-term historical data,we construct an online service cache prediction model,and propose a service cache decision algorithm with the goal of caching the most popular local services in the future.By analyzing the local service ratio and average service response time,we verified the effectiveness of the proposed mobile-aware service allocation and caching strategy.(3)For the impact of cache replacement on the computation offloading strategy,with the goal of maximizing the value of local data cache,we study the value-driven cache replacement problem and propose a value-driven cache replacement algorithm.Firstly,by introducing the priority of the corresponding computing task,we more accurately define the cache value of the data.Then,with the goal of maximizing the average cache value and the number of cache items,we define the cache replacement problem as a multi-objective optimization problem and propose a cache replacement algorithm based on the ideal point method.The algorithm ensures that the data with high cache value can be cached first.Experimental results show that the value-driven cache replacement algorithm can increase the average cache value by about 22.53%.(4)For the problem of computation offloading location selection,we study the cache-aware computing offload strategy.With the goal of minimizing the equivalent weighted response time of all tasks,we formally define the problem of cache-aware calculation offload location selection.Then,we propose an offline optimal solution based on the transportation problem and transform it into an online solution.It can choose a suitable location according to the current cache content to offload the computing task.Experimental results show that the proposed online computing offload strategy can approximate the global optimal solution,and its weighted response time is reduced by 26.13%on average.