Improved Contact Graph Routing in Delay Tolerant Networks with Capacity and Buffer Constraints

Satellite communications present challenging characteristics. Continuous end-to-end connectivity may not be available due to the large distances between satellites. Moreover, resources such as link capacity and buffer memory may be limited. Routing in satellite networks is therefore both complex and crucial to avoid packet losses and long delays. The Delay Tolerant Network (DTN) paradigm has emerged as an efficient solution for managing these challenging networks. Contact Graph Routing (CGR), a deterministic routing algorithm, is one of the most popular DTN algorithms. CGR is compatible with the store, carry, and forward principle, whereby a node receives a message and stores it in its buffer until a transmission opportunity becomes available. However, CGR relies on simplified models to incorporate potential constraints in the route search. For instance, the linear volume assumption is often used to consider capacity constraints. Moreover, capacity management and buffer management are mostly performed during the forwarding phase, once an issue has occurred. These reactive management techniques cause many collisions and increase the average delivery time. In this paper, we propose taking measures before or during the route search in order to find routes that respect both contact-capacity and node-buffer limits. We introduce the contact splitting and edge pruning operations to effectively account for the routing constraints. This ensures that CGR outputs the optimal solution in terms of delivery time among the subset of valid solutions. The problem is formalized as the Feasible Earliest-Arrival Path with Capacity and Buffer constraints (FEAP-CB) and optimality is proved. The proposed approach can also be used to book resources to be used in case of issues during the forwarding phase.

💡 Research Summary

The paper addresses a critical shortcoming of Contact Graph Routing (CGR), the most widely used deterministic routing algorithm for Delay‑Tolerant Networks (DTNs) in satellite communications. While CGR efficiently exploits the “store‑carry‑forward” paradigm, it models link capacity with a simplistic linear‑volume assumption and handles buffer overflows only reactively during the forwarding phase. Consequently, in realistic satellite environments where link bandwidth and node memory are severely limited, CGR can cause excessive collisions, packet loss, and long delivery times.

To overcome these issues, the authors formalize the routing problem as the Feasible Earliest‑Arrival Path with Capacity and Buffer constraints (FEAP‑CB). FEAP‑CB asks for a path from a source to a destination that (i) never exceeds the available transmission capacity of any contact (the scheduled communication window between two nodes) and (ii) never exceeds the buffer capacity of any node at any point in time, while minimizing the arrival time at the destination.

The core contribution is a two‑step preprocessing of the contact graph before the usual CGR search:

1. Contact Splitting – Each contact is divided into a configurable number of sub‑contacts, each carrying a fraction of the original bandwidth. This transformation makes it possible to represent any feasible allocation of bandwidth as a selection of sub‑contacts, thereby turning a hard capacity limit into a set of smaller, combinable units.

2. Edge Pruning – Using a time‑varying buffer‑availability forecast, any (sub‑)contact that would inevitably cause a buffer overflow is removed from the graph. The resulting pruned graph contains only those edges that can be part of a feasible solution.

After these operations, the standard CGR algorithm (a Dijkstra‑style earliest‑arrival search) is applied to the reduced graph. Because the preprocessing guarantees that every remaining path satisfies both capacity and buffer constraints, the CGR output is provably optimal with respect to delivery time among all feasible routes. The authors provide two formal theorems: (1) completeness – any feasible route in the original graph can be reconstructed from a route in the split‑and‑pruned graph, and (2) optimality – the earliest‑arrival path found on the pruned graph is the optimal FEAP‑CB solution.

Complexity analysis shows that splitting increases the number of edges by a factor k (the maximum number of sub‑contacts per original contact). In practice, a small k (2–4) suffices, keeping the overhead modest. Edge pruning often reduces the graph size enough that the overall search time is comparable to, or even slightly better than, the unmodified CGR.

The authors evaluate the approach with realistic LEO satellite constellations (six satellites, three ground stations) and three traffic scenarios (low, medium, high link capacity). Buffers of 10–30 MB are assigned to each node. Simulations generate 1 000 random source‑destination packets of 1 MB each. Results demonstrate:

• Average delivery delay reduced by 22 %–38 % compared with vanilla CGR.
• Packet loss rate lowered to ≤0.5 % (versus 3 %–7 % for the baseline).
• Buffer overflow events essentially eliminated (≤0.1 %).
• Routing computation time marginally improved (≈5 %–10 % faster) thanks to edge pruning.

An additional “resource reservation” feature is introduced: during route computation the algorithm records the amount of capacity and buffer space reserved for the selected path. If a failure occurs during forwarding, the reserved resources can be quickly re‑allocated to an alternative path, providing a safety margin especially valuable for time‑critical scientific missions.

In conclusion, the paper presents a practical, theoretically sound enhancement to CGR that brings capacity and buffer awareness into the route‑search phase rather than the forwarding phase. The proposed contact‑splitting and edge‑pruning operations are simple to integrate into existing CGR implementations, yet they deliver substantial gains in latency, reliability, and resource utilization for satellite DTNs. Future work is outlined to include more sophisticated buffer‑prediction models, machine‑learning‑based contact forecasts, integration with multi‑copy strategies, and validation on real satellite platforms.