| Edge computing is a growing architecture where the cloud computing resources and services are extended at the edge of networks.As a capable edge technology,it can be applied to wired,wireless and mobile scenarios,using hardware and software platforms,situated at network edge in the surrounding area of end users.Edge computing provides a seamless integration among numerous application service providers and network operators towards mobile users.Consequently,it has become an important part in the proposed successive 5G implementation that supports variation of innovative services and applications with ultra-low latency requirements.Besides,considering the heterogeneity of the edge computing a standardized term is introduced as Multi-access Edge Computing(MEC)also known as mobile edge computing.In the era of 5G communications,the core network traffic has increased sharply,and MEC networks will face the challenges of various 5G network service requirements such as high bandwidth,low latency,and high reliability.Due to the limited communication,computation and storage resources of MEC,under the sudden,intensive,and high-traffic service requests,MEC architecture will have problems such as excessive delay,service availability failure,and low distribution efficiency.In order to ensure network service quality,continuity and improve service efficiency,it is necessary to construct an effective collaborative service providing mechanism,encourage edge servers to participate in collaboration,integrate network service resources,and manage collaborative services adaptively.Consequently,this dissertation studies the service migration of high mobility devices at MEC,collaborative distribution strategy of MEC resources for services and service evaluation mechanism based on a unique collaborative MEC architecture organization strategy,Its main research work is as follows.(1)At first,MEC Fundamental Concepts and architecture coupled with the Literature Review of the smart IoT devices,its services,communications,and applications developments are illustrated.Moreover,there is categorized all the technical issues in the MEC architecture with a look over all the relevant and latest Key solutions.Further,it is discussed and pinpointed about nearly targeted Use cases and research challenges.Lastly,a critical analysis and comparison of this dissertation is summarized with existing collaborative technologies and service migration.This dissertation not only enables new readers to get a well understanding of state-of-the-art research literature but also delivers directions for new problem solving in MEC.(2)MEC is considered as a backbone for 5G and 6G network.So,the new generation network will require a collaborative MEC network that combines the properties of networking and computation of the whole MEC network to optimize performance for ultra-low latency and quality of services(QoS).The critical challenges are faced by the devices with mobility features,whether hand-held devices or vehicles fast movement from one edge server(ES)location to another ES that creating a non-optimal environment in the long run.While,the existing network technologies and approaches of service migration during mobility cannot fulfill the time complexity requirements in the large scale MEC.Hence,to maintain service continuity and to avoid service disruptions,a dynamic and distributed traffic steering system at MEC needs to design for service migration of time sensitive and real-time applications.Using the concept of different types of nodes(i.e.ES and router etc.),there can be resolved the scalability problem of a large MEC network with distributed MEC network.Hence,an updated dynamic and immutable BFS algorithm named ANBD is experimented to bounds the path searching procedure with very lesser time complexity with a filter strategy based on network distance that eliminate the non-related ESs or other network elements.With this decentralized framework of MEC,the two algorithms named MDSPS and MDMPS are proposed for dynamic path selection during service migration to minimize data transferring time.This proposed dynamic traffic steering system works under two most important metric measurements(time delay and available bandwidth).Meanwhile,the experimental results validate the effectiveness of dynamic and adaptive path searching in proposed MEC network in contrast of benchmarks that significantly outperforms other existing scheduling approaches with 35%70%efficiency in QOS.(3)In the recent development of the global 5G technology MEC enables low cost,optimal power consumption,more reliable and safer characteristics but the geographical distribution and user mobility constraints can make a tidal discrepancy,especially in metropolitan city area.In addition,during the mass migration period(morning peak hours and evening peak hours),network traffic in some geographical areas increases or decreases sharply.Thus,these massive migration periods cause traffic congestion and service blocking in high-density areas instead of free resources in a low-density area.Collaborative MEC spaces formed from the MEC server’s collaborative computation can efficiently improve the quality of experience..On the other hand,considering the tidal traffic problem,there is evaluated and build some mathematical models about whether or not the tasks can be offloaded from numerous IoT devices to MEC servers.Meanwhile,this contribution proposes a network traffic-aware service allocation and reallocation algorithms based-on dual-layered(centralized and distributed)dynamic traffic steering by extending the ANBD and MCNB algorithms.Besides,the proposed framework may also decide whether or not the service reallocation or migration is required.The main contribution of this research is about to minimize the energy consumption and time complexity during service migration with guaranteed service level agreement(SLA).The proposed approach signify the price and priority control regulatory variables for latency and time sensitive services to avoid the intrusion between users or services as guaranteed SLA and QoE.Uncertain requests of users for services,by estimated probability distribution,are used,to simulate the more realistic environment.Then,a dynamic Service Reallocation and Resource Management(DSRRM)algorithm is introduced to reallocate the multiple services,in a nut shell,among collaborative network spaces.Consequently,using the optimized regulatory variables of operational cost and energy consumption,the QoS is achieved with guaranteed SLA.Finally,the simulation results demonstrate that the proposed algorithm can achieve 26%of global cost,less than the benchmarks,with the implementation of real base-stations data set in Shanghai metropolitan area.(4)In real,the existing MEC network structure is complex,the flexibility is poor,the operation and maintenance cost is high which advance the challenges as the long development cycle of services in 5G and near future 6G.Therefore,a fundamental contribution is needed to devise as design and experimental analysis of the unified MEC slicing structure supporting to 5G communication requirements that allows the service providers to instantiate heterogeneous and collaborative services as a sliced MEC network.Using the unified and dynamic MEC coordinator,it can be compared a centralized MEC slicing framework through the extension of decentralized MEC based on ANDB algorithm for 5G applications,in this research.The main benefit is that it can provide efficient,fast,and flexible deployment for joint MEC slicing and networking.So,a threetier architecture of slices MEC is introduced for the fully optimal utilization of resources in a large scale MEC network.Consequently,there is modeled and discussed the MEC resource coupling relationships throughout computation,storage and access network,resources at each MEC node.To achieve an aggregated combination between MEC network resources,a resource coupling as an increasing linear function is made.While,the resources of diverse types are needed to define the combination of resources of different types on a specific MEC server through a one eye design of unified MEC coordinator.The main contribution is that the guaranteed resources availability is achieved during the combination of storage,computation and network resource demand.Experimental results validate the effectiveness of physical resource slicing with collateral functions where sliced MEC framework significantly outperforms than the benchmark algorithm in respect of the efficiency with guaranteed resource utilizing.Altogether,this dissertation focuses the service management in heterogeneous MEC and proposed not only collaborative service quality and management,but also designs some distributed mobility-aware service placement or migration algorithms.Based on the interactive characteristics of immutable states of MEC,both unified and distributed MEC collaborative service pool is proposed for dynamic service management,which shows high-performance improvement within a wide range of MEC sizes.Though,notable solutions exist for certain problems through the features of the proposed algorithms such as low latency,low energy consumption,migration cost minimization,management of MEC computation resources,and high efficiency with low overhead,make it suitable for implementation in a practical view. |
PDF Full Text Request