Routing and Resource Allocation for Service Function Chain in Service-Oriented Network

Service function chain (SFC) and in-network computing have become popular service provision approaches in 5G and next-generation communication networks. Since different service functions may change the volume of processed traffic in different ways, inappropriate resource allocation will lead to resource wastage and congestion. In this paper, we study the traffic change effects of service nodes with in-network computing and propose a software defined network based SFC routing and resource allocation scheme. With the objective of minimizing the difference between the service delay of adjacent function pairs in SFC and the corresponding expected delay, an optimization problem is established. Due to the coupling of computing resource provision and traffic engineering, and the non-convexity of the objective function and constraints, the problem becomes difficult to solve in practice. Therefore, we first transform the problem into a convex optimization problem using linear relaxation and variable substitution. Using the dual decomposition method, we decouple the different sets of variables. With this decoupling, the network controller can efficiently design the users' service nodes and traffic engineering. Finally, we use a rounding method to obtain a feasible solution set of the problem performed by the service nodes. Extensive simulations are performed under different system settings to verify the effectiveness of the scheme.