The Service Function Chain (SFC) is composed of multiple Service Functions (SFs) that have distinct structures and offer specific services to users. Availability is a critical metric for evaluating the SFC’s ability to deliver services. However, due to resource limitations and the dynamic nature of traffic flow, which necessitates the elastic deployment of SFCs to maintain service quality, there is a requirement to investigate how to evaluate the availability of SFCs under elastic deployment methods. In this paper, the fundamental concept and premises of the basic elasticity strategy for elastic SFC are analyzed at first. Then, utilizing Colored Petri Nets (CPNs), SFC models are proposed with different structural compositions under various elasticity strategies. These models include the serial structure, the parallel ”and” structure, and the parallel ”or” structure, and they are divided into three layers to simulate the data packets processing process of SFC, taking into account the element of time to get the time of the running time and elasticity scaling time. Subsequently, the availability of elastic SFCs is analyzed and evaluated by incorporating a mathematical model that considers elasticity. Availability values for elastic SFCs under different strategies and structures are derived through experiments. The results demonstrate that elasticity implementation has a significant impact on SFC availability. Factors such as the SFC structure, dynamic SFC deployment methods, execution time of the elastic strategies, and interval time between elastic strategy execution all influence the availability of elastic SFCs. Finally, sensitivity analyses are conducted to examine the factors that affect the availability of elastic SFCs. Ranges of important parameter values needed to ensure different levels of availability are provided. These findings shed light on the key considerations for achieving resilient SFC availability.

To efficiently meet the flexible service requests in Cloud Computing Data Centers (CloudCDC), the integrated construction of Service Function Chaining (SFC) and business system is researched based on Colored Petri Net (CPN). The general composition of multiple Service Functions (SFs) is studied at first. After analyzing the possible conflicts among the SFs that compose SFC, a novel method of SFC construction is proposed. Besides, to prove the correctness of the SFC construction, a method for verifying the constructed SFC before deployment is designed which composes the CPN model of object-oriented technology to observe the local behavior and global interaction of SFC respectively.

The service requirements from the customer are getting more and more complicated, and the users' demands for continuous uptime availability guaranteed in SLA become more and more strict. Under such circumstances, Service Function Chaining (SFC) requires more flexibility and scalability. Therefore researchers and developers have proposed plenty of complex structured solutions for the provisioning of SFC in the network. In this paper, to specify and configure these complex structured SFC, the concept of Path Adjustable SFC (P A-SFC) is proposed. After analyzing its possible traffic steering, two forwarding modes - load balancing mode and parallel mode are designed. Then the PA-SFC is modeled using YANG data modeling language. With the proposed approach, the SFC can be adjusted into a parallel chain or multiple loads balancing chains automatically under the control of the SDN controller.

Graph convolution neural network is one of the hot demanding research areas in the last few years. Due to the issues of data irregularity and computation complexity driven by typical GNN networks, we propose a spatial temporal graph convolution network for action recognition system on FPGA(STARS). STARS has redesigned several computing kernels based on the original layers of ST-GCN and adopted specific algorithm with optimization strategies for different kernels. To optimize the performance of accelerator, the ping-pong buffers for data transmission and dynamic quantification for model inference are implemented. The effectiveness of STARS driven accelerator is verified on Xilinx Pynq-Z1 prototyping board.

Service applications are applicable to provide services for requests of users from network. Due to the fact that they have to endure a big number of concurrent requests, the performance of service applications running under specific arrival rate of requests should be assessed. To measure the performance of a service application, multi-party testing context is needed to simulate a number of concurrent requests and collect the responses. TTCN-3 is a test description language; it provides basic language elements for multi-party testing context that can be used in performance tests. This paper proposes a general approach of using TTCN-3 in multi-party performance testing service application. To this aim, a model of service application is presented, and performance testing framework for service applications is discussed. This testing framework is realized for a typical application by developing a reusable TTCN-3 abstract test suite.