High Capacity and Resilient Large-Scale Deterministic IP Networks

With 5G networking, deterministic guarantees are emerging as a key enabler. In this context, we present a scalable architecture for Large-scale Deterministic IP Networks (LDN) that meets end-to-end latency and jitter bounds. This work extends the original LDN (Liu et al. in IFIP Networking, 2021) architecture, where flows are shaped at ingress gateways and scheduled for transmission at each link using an asynchronous and cyclic opening of gate-controlled queues. To further optimize the utilization of bandwidth and accept more traffic, our Advanced-LDN (A-LDN) architecture introduces a new shaping mechanism based on transmission patterns. To protect flows against failures, we also implement Frame Replication and Elimination for Reliability (FRER). To do so, we leverage on the A-LDN mechanism for traffic scheduling and activate it at core nodes, i.e., adding the possibility for additional shifts inside core nodes, so that replicas can arrive at destination at the same time. In any case, the network remains stateless (no per-flow states at core nodes), ensuring scalability over large-scale networks. For the control plane, we present variants of a column generation algorithm to quickly take admission control decisions and maximize traffic acceptance. For a set of flows, it determines acceptance and selects the best shaping parameters, routing policy, transmission patterns, and scheduling. Through numerical results and packet level simulations in OMNeT++, we demonstrate that our A-LDN architecture with transmission patterns improves traffic acceptance by up to 67% compared to the original LDN architecture. We also demonstrate that FRER can be efficiently supported at the cost of some extra bandwidth utilization.