How Will 7G Travel on Multiple Paths Simultaneously?

7G is the next mobile network generation, set to combine satellites, airships and ground stations. Imagine sending a video to a friend across the globe using not one but many routes at the same time. This is the idea of multi‑path transmission: simultaneous paths boost speed and data reliability.

Traditional technologies like MPTCP or MPQUIC struggle with 7G's ultra‑low latency and smart route selection needs. The Information‑Centric Network (ICN) steps in, focusing on content rather than addresses. ICN is highly mobile and uses bandwidth efficiently, ideal for such a diverse environment.

The AIICN project merges ICN with artificial intelligence to decide in real time which route is best. Researchers modeled the problem with a mixed‑integer linear programming approach, solving path‑conflict issues. Tests show AIICN raises throughput by about 21% compared to traditional methods, while keeping complexity low and converging fast.

In practice, 7G will deliver ultra‑smooth streaming, lag‑free online gaming and reliable connections even in remote areas, thanks to this new AI‑driven multi‑path transmission based on ICN.

The AIICN model was formulated as a mixed‑integer linear programming (MILP) problem. In this formulation, binary variables represent the activation or deactivation of a specific path between source and destination nodes, while continuous variables describe the amount of data flow assigned to each path. The objective is to maximize total throughput, subject to constraints on bandwidth capacity, maximum allowable latency, and protocol compatibility.

MPTCP and MPQUIC limitations stem from their design based on traditional end‑to‑end connections, which do not account for the dynamic nature of 7G nodes such as orbiting satellites or moving airships. These technologies require complex handshake negotiations and maintain connection states that increase latency.

ICN, on the other hand, uses a content‑naming model (Content‑Name) that allows data to be addressed independently of the node's physical location. This facilitates mobility and content replication across multiple paths. Artificial intelligence is employed to predict path quality in real time, leveraging network metrics like jitter, packet loss, and throughput variations. A reinforcement learning (RL) algorithm evaluates rewards associated with each routing decision, updating the path‑selection policy.

In the simulation phase, scenarios with 100 nodes distributed across three layers—LEO satellites, low‑altitude airships, and terrestrial base stations—were created. The MILP model was solved with a commercial solver (Gurobi) and compared to heuristic solutions based on MPTCP. Results show convergence in fewer than 50 iterations, with an average 21% throughput improvement and a 15% latency reduction compared to traditional methods. Complexity analysis indicates that, thanks to Benders decomposition techniques, computation time remains acceptable even for large‑scale networks.

In summary, AIICN merges ICN's flexibility with AI's predictive power, solving the multi‑path routing problem in highly dynamic 7G environments while ensuring bandwidth efficiency and low latency.