We will use Multipath Transport protocols for communication between a sender and receiver. This will let us use multiple network paths, including wireless channels, to increase bandwidth and resilience against disruptions. We’ll use existing multipath transport protocol implementations on end systems, like user devices and application servers. The Multipath application will work with network functions (e.g., rApps and the Multi-Operator SMO) to learn about available paths in the 4G/5G network. End systems will use these paths based on their characteristics. The rApp learns from the MO-SMO and AI Planner about the available paths, aiming to meet performance and resilience needs.

Multipath TCP (MPTCP) and MPQUIC (a multipath version of QUIC) are key protocols. MPQUIC, implemented in user-space, is easier to enhance for testing. We plan to innovate in scheduling, reliable delivery, and congestion control. Given the variability of high-speed wireless links like mmWave, it’s crucial to adapt to changes in link characteristics quickly. We aim to predict these characteristics in real-time to utilize available bandwidth effectively. A data-driven model for bandwidth, delay, and loss will guide packet allocation among different links.

For resilience, we’ll add redundancy during transmission using source and network coding. This redundancy will be proactive (sent with data packets) and reactive (sent when proactive measures are insufficient). This approach helps handle packet loss and maintains performance.

We’ve been testing Multipath Transport with simulations and a simple testbed. We will integrate the scheduler and redundancy mechanisms into MPQUIC. The main task is developing the rApp to integrate with the MO-SMO and AI Planner, ensuring the user equipment uses the best available paths to meet performance requirements.

We will also use multipath connections to securely distribute symmetric keys via secret sharing. Using multiple paths, we can distribute shares of the key, making it harder for attackers to intercept all necessary pieces to reconstruct the key.
Role of the Resilience rApp and Multipath Controller

The AI Planner takes user input on service requirements (like video, voice, or data transfer needs) and assesses if the network can meet these needs. It communicates these requirements to the MO-SMO and the Resilience rApp, which evaluates the need for multipath communication and resource selection. The AI Planner and MO-SMO translate user requirements into network-level Quality of Service (QoS) capabilities.

The Resilience rApp determines which paths will provide the required QoS for user devices (UEs) by interacting with the MO-SMO’s Inventory and Topology service. This service knows the UEs, their links (e.g., 5G, LTE, WiFi, satellite), and the QoS characteristics of available network slices.

The Resilience rApp communicates with the Multipath Controller, which tells the UE which interfaces (and paths) to use to connect to remote nodes. This ensures that the selected paths meet the desired Quality of Experience (QoE) for the user. In the future, the Multipath Controller will also consider routing to optimize path selection further.
Summary

Our approach leverages Multipath Transport protocols to enhance communication resilience and performance by using multiple network paths. We will innovate in scheduling, delivery, and congestion control to adapt to high-speed wireless link variability. Our system integrates with AI and network functions to ensure user requirements are met, even in challenging conditions, using both proactive and reactive redundancy strategies. Additionally, we will use multipath connections for secure key distribution, enhancing overall system security.