Research On Service Adaptation Mechanisms In Smart Collaborative Integrated Networking

As the world's largest information sharing platform, the Internet has play an irreplaceable role in many areas such as science and technology, national economy, and people's livelihood. However, due to its inherent shortcomings in design,the current Internet cannot well support various requirements of today's users and operators,and has exposed many serious problems hindering its sustainable developments.Consequently, designing the future network architectures has received great attentions around the world. Nevertheless, such the research field is still in its early stage, and related mechanisms are far from perfect. As the potential solutions, the smart collaborative networking and integrated identifier networking have promising prospects for their significant enhancements of network openness, flexibility, controllability, and efficiency.In this dissertation, based on the extensive investigations of future network researches, we concentrate on the two key issues of the smart collaborative integrated networking (the combination of smart collaborative networking and integrated identifier networking), the specification and improvement of architectural details, and the design of service adaption mechanisms. As a result, the main contributions of this dissertation are four-fold shown as follows.(1) To solve the existing problems of the current Internet in integration of heterogeneous networks, mobility support, and transmission efficiency, we propose a network construction scheme based on the smart collaborative integrated networking,which re-defines and re-designs the network naming mechanism, the components and their functions, and the registration and communication process. To evaluate the performance of the proposed scheme, we build a proof-of-concept prototype and conduct extensive experiments. The related results confirm the feasibility of the proposed scheme and its advantages. For example, satellite and terrestrial networks are able to be integrated effectively, and endpoint handover delay has reduced from the order of seconds to milliseconds.(2) To achieve efficient service content adaption, we propose two caching permission policies with time complexity at O(1), for routers to identify popular contents and cache them on-demand. The purposes are to improve user experience,reduce server burden and network traffic. To assess the performance of the proposed two policies, extensive simulations are performed, with another 7 related policies implemented as comparisons. The results show that the proposed two policies outperform others. For example, in terms of content-access distances, compared with Leave Copy Everywhere (LCE) used by Content-Centric Networking (CCN), the proposed two policies save over 17% number of hops.(3) For better service content adaption, we further propose a cache consistency mechanism to guarantee the freshness of cached contents in routers, which balances the trade between the consistency strength and related costs. Besides, we improve another two cache consistency mechanisms used in the web caching, namely Polling-Every-Time (PET) and Time-To-Live (TTL), to be suitable for router caching networks. To evaluate their performance, we firstly analyze the costs of these mechanisms, and then conduct extensive simulations. The relevant results show that the proposed mechanism reduces the content-access distances sharply compared to the improved PET, and the user content staleness rate remarkably compared to the improved TTL.(4) To realize effective service function adaption, we firstly propose a service-oriented communication framework for network function chains support in the proposed smart collaborative integrated networking. Then, we further propose an energy-efficient optimization model and a related heuristic algorithm for virtual service function chain embedding. Finally, we implement a proof-of-concept prototype, and prove the feasibility and availability of the proposed framework. In addition, we perform extensive simulations and confirm the advantages of the proposed optimization model and corresponding heuristic algorithm in energy conservation, with respect to the number of active virtual machines and remaining node resources of substrate networks.By these research works, we further enrich and improve the smart collarative integrated networking, providing a new solution for future networks.