Sina Keshvadi | Searcharxiv

arXiv2026

Performance Analysis of Dynamic Equilibria in Joint Path Selection and Congestion Control in Path-Aware Networks

Path-aware networking (PAN) architectures, such as SCION and emerging LEO constellations, expose tens to hundreds of verifiable paths to endpoints. When multipath protocols like MPTCP and MPQUIC greedily exploit this diversity, uncoordinated migration can induce persistent, high-amplitude load oscillations. Although this instability is well-known, its quantitative performance impact remains poorly understood. In this paper, we apply a discrete-time axiomatic framework to the joint dynamics of loss-based congestion control and greedy path selection. By deriving the system's dynamic equilibria (stable periodic oscillations), we prove a fundamental trade-off: high Responsiveness improves Fairness but necessarily degrades Efficiency and Convergence. Conversely, we demonstrate that Efficiency, Convergence, and Loss Avoidance are simultaneously achievable at a critical lossless operating point. Furthermore, we find that while migration de-synchronizes traffic in high-diversity environments, realistic limited-visibility constraints transform coherent oscillations into persistent spatial load imbalance, rather than eliminating instability entirely. These results yield concrete design guidelines for robust multipath transport over the future path-aware Internet.

cs.NI↗

arXiv2025

SCION Path Performance Toolkit and Benchmark for Advancing Machine Learning in Next-Generation Networks: ScionPathML

Path-aware networks promise enhanced performance and resilience through multipath transport, but a lack of empirical data on their real-world dynamics hinders the design of effective protocols. This paper presents a longitudinal measurement study of the SCION architecture on the global SCIONLab testbed, characterizing the path stability, diversity, and performance crucial for protocols like Multipath QUIC (MPQUIC). Our measurements reveal a dynamic environment, with significant control-plane churn and short path lifetimes in parts of the testbed. We identify and characterize path discrepancy, a phenomenon where routing policies create asymmetric path availability between endpoints. Furthermore, we observe a performance trade-off where concurrent multipath transmissions can improve aggregate throughput but may degrade the latency and reliability of individual paths. These findings demonstrate that protocols such as MPQUIC should explicitly account for high churn and path asymmetry, challenging common assumptions in multipath protocol design.

cs.NI↗

arXiv2025

An Axiomatic Analysis of Path Selection Strategies for Multipath Transport in Path-Aware Networks

Path-aware networking architectures like SCION provide end-hosts with explicit control over inter-domain routing, while multipath transport protocols like MPTCP and MPQUIC enable the concurrent use of multiple paths. This combination promises significant gains in performance and policy enforcement, but it also creates a stark trade-off between individual performance optimization and overall network stability. This paper quantifies this trade-off through a rigorous axiomatic analysis. We evaluate a spectrum of algorithms, from greedy (Min-RTT) and cooperative (Round-Robin) to hybrid approaches (Epsilon-Greedy), against axioms of Efficiency, Loss Avoidance, Stability, and Fairness in a simulated path-aware environment. Our simulations reveal that purely greedy strategies, while efficient under low contention, induce catastrophic packet loss, increasing by over >18,000% as the number of competing agents grow, due to herd effects that cause severe network instability. Conversely, cooperative strategies ensure fairness and stability but at the cost of underutilizing high-capacity paths. Crucially, we demonstrate that hybrid strategies resolve this conflict. The Epsilon-Greedy algorithm, for instance, achieves the highest efficiency of all tested strategies in high-contention scenarios while mitigating the instability inherent to the greedy approach. Our axiomatic analysis suggests that tunable, hybrid algorithms are essential for designing robust and high-performance path selection mechanisms for next-generation networks.

cs.NI↗

arXiv2025

Path Dynamics in a Deployed Path-Aware Network: A Measurement Study of SCIONLab

Path-aware networks promise enhanced performance and resilience through multipath transport, but a lack of empirical data on their real-world dynamics hinders the design of effective protocols. This paper presents a longitudinal measurement study of the SCION architecture on the global SCIONLab testbed, characterizing the path stability, diversity, and performance crucial for protocols like Multipath QUIC (MPQUIC). Our measurements reveal a dynamic environment, with significant control-plane churn and short path lifetimes in parts of the testbed. We identify and characterize path discrepancy, a phenomenon where routing policies create asymmetric path availability between endpoints. Furthermore, we observe a performance trade-off where concurrent multipath transmissions can improve aggregate throughput but may degrade the latency and reliability of individual paths. These findings demonstrate that protocols such as MPQUIC should explicitly account for high churn and path asymmetry, challenging common assumptions in multipath protocol design.

cs.NI↗