Backpressure with Adaptive Redundancy (BWAR)

Backpressure scheduling and routing, in which packets are preferentially transmitted over links with high queue differentials, offers the promise of throughput-optimal operation for a wide range of communication networks. However, when the traffic load is low, due to the corresponding low queue occupancy, backpressure scheduling/routing experiences long delays. This is particularly of concern in intermittent encounter-based mobile networks which are already delay-limited due to the sparse and highly dynamic network connectivity. While state of the art mechanisms for such networks have proposed the use of redundant transmissions to improve delay, they do not work well when the traffic load is high. We propose in this paper a novel hybrid approach that we refer to as backpressure with adaptive redundancy (BWAR), which provides the best of both worlds. This approach is highly robust and distributed and does not require any prior knowledge of network load conditions. We evaluate BWAR through both mathematical analysis and simulations based on cell-partitioned model. We prove theoretically that BWAR does not perform worse than traditional backpressure in terms of the maximum throughput, while yielding a better delay bound. The simulations confirm that BWAR outperforms traditional backpressure at low load, while outperforming a state of the art encounter-routing scheme (Spray and Wait) at high load.

💡 Research Summary

The paper addresses a well‑known drawback of backpressure‑based routing: when traffic load is low, most queues are empty or near empty, causing packets to wander through the network for a long time before reaching their destination. This problem is especially acute in intermittently connected mobile networks (delay‑tolerant networks, DTNs), where contacts between nodes are sparse and the overall system is already delay‑limited. Existing DTN solutions such as Spray‑and‑Wait mitigate delay by creating multiple copies of each packet, but they do so with a fixed replication factor; consequently, when the load becomes moderate or high the extra copies cause congestion and increase delay.

The authors propose a hybrid scheme called Backpressure with Adaptive Redundancy (BWAR). Each node maintains the usual per‑destination queue Q (the “original” queue) and, in addition, a duplicate buffer D of limited size. When a node transmits a packet and its original queue occupancy is below a pre‑defined threshold q_th, the transmitted packet is also placed in the duplicate buffer (provided the buffer is not full). Duplicate packets are never removed from D by transmission; they are only removed when the destination acknowledges receipt or when a global timeout P expires. Moreover, when a link is scheduled, the node first exhausts all packets in Q before sending any from D, guaranteeing that duplicates are used only when the original queue is empty. This policy ensures that in low‑load regimes many copies of a packet are scattered throughout the network, dramatically increasing the chance that a copy meets the destination during a random encounter, while in high‑load regimes duplicates are rarely created and the algorithm behaves exactly like classic backpressure.

To incorporate the duplicate information into routing decisions, the authors modify the backpressure weight for each commodity c on link (i, j) as follows:

\