Abstract: Compute first networking is an emerging paradigm to meet the ever-increasing computation demands. an important part of the future national information infrastructure. However, current challenges, such as security issues, low resources utilization, and heterogeneity, invoke an urgent need for effective collaborative mode of value chain for computing first networking. Therefore, it is of significant importance to seek a collaborative method of computing and networking resources scheduling value chain to upgrade digital infrastructure. Firstly, the proposed framework contains a new computing and networking mixed scheduling algorithm to rapidly converge into a valid solution. Secondly, it constructs computation models and transition functions to model the computing networking and resources scheduling problems, and builds a greedy-based scheduling policy to dynamically make decisions for different networking states. Thirdly, the proposed framework recursively computes environment state in the form of a sequence of states according to variables and environment via transitions functions and computation models. Next, the loss of the state trace is computed in the term of a compound loss function. Gradient descent is utilized to minimize the loss by updating variables, reschedule and compute new a state trace based on updated variables, until the loss converges into a valid plan for the planning problem. At last, the above modules compose a new value chain collaborative design method in a simulation environment. The experimental results show that the proposed method is both effective and efficient in resources scheduling.

Key words: compute first networkingvalue chain, collaborative mode, resource scheduling, mobile-edge computing