Modeling And Optimization Of Wireless Communication Networks Based On Mobility-aware Caching

With rapid emergence of new services and application scenarios,explosive growth in mobile data traffic and massive device connectivity are becoming two main challenges for wireless networks.To deal with these challenges,5G networks use a lot of advanced technologies,e.g.e,wireless caching technology.With caching,the networks place the contents into the edge devices(including base stations and user equipments,but this paper considers only caching in user equipments)in advance.When massive users request the same content,they can obtain the content from the edge devices so as to avoid the network repetitively deal with the request.Therefore,wireless caching not only reduce the downloaded delay,but also alleviate reduce the burden of backhaul.This paper mainly investigates user equipments in mobility scenarios,and makes the best of cooperative communications between nodes to improve caching performance.Firstly,a mobility-aware based cost-optimal caching problem(COCP)of file recovery is modeled.COCP analyzes the impact of user mobility,cache size,and total number of encoded segments are all taken into account.The hardness of the problem is proved.For solving COCP,a lower-bounding approximative function of the objective function of COCP is derived.By the approximative function,an approximation of COCP(ACOCP)approach achieving linearization is constructed.For small-scale and medium-scale system scenarios,ACOCP approach can obtain a lower bound of global optimum and a sub-optimal solution of COCP.In addition,a low complexity mobility aware multi-user algorithm(MAMU),achieving the tradeoff between complexity and accuracy.The evaluation results manifest that the ACOCP approach and the MAMU algorithm outperform conventional caching algorithms.With the increase of mobility speed,mobility first takes positive gains,and then negative gains.Secondly,as the existing works based on mobility-aware caching ignore the relationship between delay and offloading gain,a delay-optimal caching problem(DOCP)satisfying the given offloading gain is modelled.Solving DOCP is equivalent to solving a network load ratio(NLR)optimal caching problem(NOCP)with performance guarantee.A bisection search approach and an approximative NOCP problem(ANOCP)approach are developed to obtain the low bound of global optimum of DOCP.Based on the low bound,an effective search algorithm(ESA)is proposed to obtain the sub-optimal of DOCP.In ESA,to reduce the complexity of integer programming algorithm in the ANOCP approach,a relax-rounding algorithm is proposed.The evaluation results show that compared to conventional caching schemes,the proposed algorithms can realize the given offloading target in a shorter delay duration.In addition,solving integer programming(ILP)realize the tradeoff between complexity and accuracy.Thirdly,the existing works on mobility-aware caching ignored the impact of the caching time duration of contents on caching optimization.Based on this observation,a retentionaware caching problem(RACP)in vehicular networks is modelled.The hardness of RACP,in terms of the computational complexity,is proved.An optimal dynamic programming(DP)algorithm is derived to achieve RACP's global optimum for symmetric scenarios where the helpers' caching capacities as well as the contact rates are uniform.To tackle general system scenarios,a low-complexity retention aware multi-helper caching algorithm(RAMA)is proposed.The evaluation results show that in symmetric scenarios,with the increase of the weight of the storage cost and the total number of slots,the gap of the solutions obtained from RAMA to the global optimum gradually decreases.In general scenarios,for the low mobility scenarios,the popularity-based caching scheme achieves excellent performance.Besides,the proposed algorithms outperform conventional caching algorithms.Finally,the existing works based on mobility-aware caching consider only the scenarios that the popularity of contents are fixed.This paper models a dynamic caching optimization problem in vehicular networks(DCVN)to investigate the impact of the contents' popularity changes over time on caching optimization.The hardness of DCVN,in terms of the computational complexity,is proved.For solving DCVN,the DCVN's form is equivalently transformed into a linear integer programming form.Thus,the optimum of DCVN can be derived in small-scale and medium-scale system scenarios.In addition,a low complexity greedy algorithm is proposed to obtain a sub-optimal solution The evaluation results show that compared to the global optimum,when the cost of updating unit data is low,frequent updating cache achieves excellent performance.The proposed greedy algorithm outperforms the popularitybased caching scheme.In addition,this paper develops an embedded Markov chain approach to analyze the signaling cost of the racking area list(TAL)-based location management scheme.The accuracy of the proposed model is validated through Mont Carlo simulation.The results manifest that the network can allocate,to a UE unit,an optimal TAL that is consistent with the UE's mobility and traffic characteristics to minimize the signaling cost of the TAL-based scheme.The allocation scheme based on the proposed model outperform the existing allocation schemes.