Dependency-aware cache optimization and offloading strategies for intelligent transportation systems

With the proliferation of structured applications in intelligent transportation systems, cloud-edge-end collaboration technology has gained widespread attention. In order to reduce the offloading delay and energy consumption of structured dependency subtasks while balancing the load of edge servers, a subtask dependency structure partitioning strategy was proposed in this paper. This policy categorizes the dependencies between subtasks into two types: serial dependencies and parallel dependencies. Based on these classifications, a popular dependency-aware cooperative caching policy (PACCS) was designed, which considers the fitness of popular subtasks with different server resource sizes. Then, we design a delay model, an energy consumption model, and an edge server load balancing model to achieve a multi-objective optimization that integrates system delay, energy consumption, and edge server load balancing using the improved NSGA-III algorithm (S-NSGA-III). Simulation experiments show that under the same experimental conditions, the integrated cost of the S-NSGA-III adaptive optimization scheme proposed in this paper is 13.0% lower than that of the NSGA-II scheme, 12.2% lower than that of the NSGA-III scheme, and 16.5% lower than that of the PeEA scheme.