The rapid emergence of mission-critical applications has shifted network requirements toward strict end-to-end deterministic delay guarantees. While packet-switching infrastructures provide flexibility and scalability, achieving predictable delay remains challenging in the presence of arrival jitter caused by traffic shaper implementations, which are used to regulate packet transmission intervals, and initial timing offsets during connection establishment. This paper presents a theoretical framework for analyzing delay performance in periodic credit-based input-queued switches under bounded arrival jitter, where the jitter does not exceed the transmission period T. We prove that the maximum queueing delay is bounded by 2T when the initial offset Φ = 0, and by 3T during the pre-connection establishment phase, where the initial phase offset between the arrival process and the credit assignment is not yet fixed. Furthermore, we show that the probability that the delay exceeds a given threshold prior to connection establishment is upper-bounded by a distribution obtained via convolution of the Φ=0 case. Numerical results validate the derived bounds and demonstrate that the framework enables reliable worst-case delay estimation without exhaustive simulations.
In multi-domain networking, virtual network function (VNF) scaling using machine learning requires an accurate prediction model while addressing privacy constraints and non-identical and independently distributed (non-IID) data across domains. Current models have used conventional federated learning (FL) methods, such as federated averaging (FedAvg), yet they suffer from degraded performance due to heterogeneous traffic patterns in multi-domain networks. However, existing studies have not addressed the impact of non-IID characteristics on FL-based VNF scaling or developed an effective solution to mitigate it. This paper proposes a metadata-clustering-driven FL method that clusters domains with different traffic patterns and trains cluster-specific models. We extract statistical, spectral, and temporal features to represent traffic disturbance. We apply principal component analysis (PCA) followed by K-means clustering to group time series. We apply FedAvg within clusters to train cluster-specific prediction models. To evaluate the performance of the proposed method, we set up a testbench to synchronize three non-IID patterns. The numerical results demonstrate that the proposed clustered FL method consistently achieves a lower mean squared error (MSE) than the FedAvg baseline across all four evaluated non-IID settings. The proposed method yields an MSE of 0.7056 (a 23.1% reduction from FedAvg’s 0.9176) under label skew, 0.3615 (a 4.6% reduction from 0.3790) under label and feature skew, 0.6958 (a 28.1% reduction from 0.9682) under label and quantity skew, and 0.3748 (a 0.7% reduction from 0.3774) under the combined skew setting. These consistent reductions in MSE demonstrate that the proposed method effectively mitigates the performance degradation typically caused by non-IID effects.
Run-Yu Wang, Eiji Oki· International Conference on...· 0 citations
Simulation results show that the CIOQ switch with VIQs effectively isolates critical flow from the impact of bursty flow originating from other input ports, thereby mitigating delay degradation caused by cross-port interference under the examined traffic conditions.
Takuto Kubo, Shingo Okada, Eiji Oki· IEEE Open Journal of the Com...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
