Abm

Addanki, Vamsi; Apostolaki, Maria; Ghobadi, Manya; Schmid, Stefan; Vanbever, Laurent

doi:10.1145/3544216.3544252

Cited by 20 publications

(4 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore any policy that satisfies (33) and ( 34) leads to minimum Davg , which is T 2µ max 4 i=1 λ i − µ , 0 . Similarly, we can show (33) and (34) together form a sufficient condition to minimize the maximum ingress delay Dmax .…”

mentioning

confidence: 82%

“…Note that we can easily derive a weighted version of ( 19) based on the node buffers that can balance the buffer utilization at each layer [34], [67], [68]. The main point we convey from this example is the high flexibility of choosing γ according to different objectives in network control.…”

Section: A Practical Extensions 1) Multi-objective Optimization With ...mentioning

confidence: 99%

“…The delay increase caused by network overload is mainly due to the queueing delay of packets, since the queue buffers are increasingly backlogged due to the overload. A common approach to reducing queueing delay is active queue management, which drops packets when overload is observed or the queueing delay exceeds a specific upper bound [32]- [34]. Alternative approaches include network calculus-based scheduling with worst-case latency guarantees [3], [35], traffic shaping and pacing [36], [37], and smart buffer design to absorb traffic spikes [38].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Queueing Delay Minimization in Overloaded Networks via Rate Control

Modiano

2022

2022 58th Annual Allerton Conference on Communication, Control, and Computing (Allerton)

View full text Add to dashboard Cite

We develop link rate control policies to minimize the queueing delay of packets in overloaded networks. We show that increasing link rates does not guarantee delay reduction during overload. We consider a fluid queueing model that facilitates explicit characterization of the queueing delay of packets, and establish explicit conditions on link rates that can minimize the average and maximum queueing delay in both single-hop and multi-stage (switching) networks. These min-delay conditions require maintaining an identical ratio between the ingress and egress rates of different nodes at the same layer of the network. We term the policies that follow these conditions rate-proportional policies. We further generalize the rate-proportional policies to queue-proportional policies, which minimize the queueing delay asymptotically based on the time-varying queue length while remaining agnostic of packet arrival rates. We validate that the proposed policies lead to minimum queueing delay under various network topologies and settings, compared with benchmarks including the backpressure policy that maximizes network throughput and the max-link-rate policy that fully utilizes bandwidth. We further remark that the explicit min-delay policy design in multi-stage networks facilitates co-optimization with other metrics, such as minimizing total bandwidth, balancing link utilization and node buffer usage. This demonstrates the wider utility of our main results in data center network optimization in practice.

show abstract

mentioning

confidence: 82%

Section: A Practical Extensions 1) Multi-objective Optimization With ...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Queueing Delay Minimization in Overloaded Networks via Rate Control

Modiano

2022

2022 58th Annual Allerton Conference on Communication, Control, and Computing (Allerton)

View full text Add to dashboard Cite

show abstract

“…Several researches have focused on the use of advanced traffic statistics for the estimation of congestion state. These traffic statistics include traffic load pattern [24], buffer size with delay [25], traffic labels [26], traffic classes (like best effort, real-time, and low latency) [27], and disturbance observer and smith predictor [28]. Some advanced techniques used reinforcement learning-based drop decisions [29], policyoriented drop probability based on delay and resource utilization [30], and flow statistics-based drop estimation [31].…”

Section: Literature Reviewmentioning

confidence: 99%

Adaptive In-Network Queue Management using Derivatives of Sojourn Time and Buffer Size

Saleh,

Shu,

Koldehofe

2024

NOMS 2024-2024 IEEE Network Operations and Management Symposium

View full text Add to dashboard Cite

Active Queue Management (AQM) algorithms are heavily used in packet processors to maintain an optimal queue size and avoid issues like Bufferbloat. Despite the remarkable performance, the traditional AQM algorithms face a major challenge of estimating the accurate queue congestion due to bursty network conditions. The major reason is the use of baseline queue statistics for congestion estimation like delay and sojourn time for Random Early Detection (RED) and Controlled Delay (CoDel), respectively. In this paper, we propose a novel dAQM algorithm that uses advanced traffic statistics like three higher-order derivatives of sojourn time and buffer size along with the baseline sojourn time and buffer size for accurate congestion estimation. dAQM adjusts its drop rate based on the continuously varying congestion to cater to the needs of bursty traffic. We simulated dAQM in ns-3 and analyzed its performance for FTP traffic by variation in traffic load and packet sizes. The results showed that dAQM provides at least 25% and 39.7% reduction in packet loss ratio and flow completion time, respectively, as compared to the traditional AQM algorithms.

show abstract