Research On Key Technologies Of Polymorphic Routing In Reconfigurable Basal Network

With the continuous development of the new generation of information technologies, such as cloud computing, Internet of things, and big data, the Internet has penetrated nearly every aspect of human life, economy and society. "Internet+industries" will deeply integrate Internet innovation into all economic and social areas, and form a new normal for broader economic development, which is based on Internet infrastructure and implementation tool. However, the IP semantic overload and poor service customization ability restrict the further development of the "Internet+". The gap between the service demand and network capacity is more significant. Therefore, how to provide a variety of network services quickly, flexibly and efficiently has become a hot and difficult point in the future network research.At present, the network routing faces two major challenges:firstly, how to adapt to the increasing network scale and improve the scalability of the routing system is the first major challenge. Secondly, how to improve the service ability of the routing system in different business characteristics and the changing quality of the service demand is the second challenge. Under this background, this dissertation analyzes and researches the key technologies of the polymorphic routing in the reconfigurable basal network, and tries to explore an effective technological path for enhancing the capacity of network interconnection and transmission.This dissertation focuses on how to improve the routing scalability and the diversification of service capacity, and proposes a polymorphic routing model in reconfigurable basal network. The basic idea behind the polymorphic routing model is decomposition and composition. Based on the polymorphic routing model, we research the mapping resolution system, the polymorphic derivation method in network nodes, and the optimal service path selection among network nodes. At last, we build a prototype system of the polymorphic routing, which can support the coexistence of a variety of network architectures and the routing systems. To summarize, the main research results of this dissertation are as follows:■ We propose a polymorphic routing model in the reconfigurable basal network. The model is based on decomposition and composition under the identifier/locator separation. In this model, the traditional routing protocols and network services are divided into the basic functional units, called routing atomic capacities. The customized polymorphic routing protocol is derived through the composition of different routing atomic capacity instances. The key of polymorphic routing model consists of three parts. The first is the description of routing service requirement, called routing service description, which expresses the requirements of applications desired by PR model. The second is base state routing which is the set of routing atomic capabilities. The third is the polymorphic derivation, which is the key to realize from the base state routing to the polymorphic routing. The model enables the network to self-adjust routing services dynamically to adapt to the different requirements of applications, supports coexistence of multiple routing modes and communication paradigms. Case studies and simulation results show that the polymorphic routing model has certain advantages in supporting content delivery, service migration and mobility, compared to the traditional network.■ Aiming at the scalability and high resolution latencies in the current existing mapping system, we present LMChord, a layered mapping system based on locality-aware DHT. To address the mismatch problem between overlay and physical network, we present the LMChord construction model, which models the LMChord construction process as a Markov decision process (MDP). Moreover, we present a Markov decision construction algorithm, which improves reinforcement learning to get the global optimal or near-optimal construction strategy. To further improve routing efficiency, we also modify the finger table to optimize the LMChord’s routing hops. We show that, besides the capability to support incremental deployment and flat identifiers, the mapping scheme is more scalable and has lower resolution latency. The evaluation also demonstrates the performance of our approach.■ Aiming at the derivation problem of the polymorphic routing protocol in network nodes, we propose a polymorphic derivation algorithm based on Markov Approximation (PDMA). The polymorphic derivation problem is mapped into a multi-dimension multi-choice knapsack problem (MMKP). To solve the combinatorial optimization problem, the problem is approximated as a Markov approximation problem by the log-sum-exp function. The optimal solution of the approximation problem is the stationary distribution for a class of time reversible Markov chains. We design a distribution algorithm to implement such a Markov chain, to optimize the network utility. The experimental results show that PDMA can effectively derive the polymorphic routing protocols and enhance the network overall utility. Comparing with the genetic algorithm, the running time is reduced by more than 20%, and the network utility at least increases by more than 5% under the same success ratio.■ We present a multiple objective and multiple constraints optimal service path selection algorithm (MOPSO) based on particle swarm optimization. Aiming at the service path selection, we define the measure metrics. Based on the metrics, we mapped the multi-objective optimal service selection problem into a particle swarm optimization model, which can speed the search using parallel search method. To solve this model, we design a suitable particle swarm optimization algorithm. Case studies and simulation results show that MOPSO can obtain the approximate optimal solution regardless of the size of problem and population. Comparing with the genetic algorithm, MOPSO has smaller oscillation and the number of iterations under the same optimal solution.■ We design and implement a polymorphic routing prototype system which can support the coexistence of different network architectures and routing protocols. The prototype system based on NetFPGA, using OpenVZ and Quagga as the core, achieves a routing solution for the diversified applications when multiple network architectures coexist. In the control plane, OpenVZ virtualizes multiple virtual containers, and Quagga deploys different routing protocols in different virtual container. In the data plane, we realize flexible and customized data packet forwarding based on the programmable NetFPGA. The tests in the real network environment verify the function and performance of the polymorphic routing prototype system.