LLM Inference Under Bursty Workload Distribution: Modifying the WAIT Algorithm

• URL: http://arxiv.org/abs/2608.06135v1
• Date: Thu, 06 Aug 2026 15:07:43 GMT
• Title: LLM Inference Under Bursty Workload Distribution: Modifying the WAIT Algorithm
• Abstract summary: We propose a lightweight extension to the state-of-the-art WAIT algorithm.<n>The proposed algorithm performs online estimation of request intensity based on observed interarrival times.
• Score: 0.8564127299257368
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large Language Models (LLMs) such as ChatGPT and Claude are widely used for information retrieval and problem-solving. Recent work has focused on improving scheduling algorithms to boost throughput while maintaining low latency. However, these approaches often assume Poisson request arrivals with constant rates - an assumption that fails to reflect the inherently bursty and dynamic nature of real-world traffic. We propose a lightweight extension to the state-of-the-art WAIT algorithm [1], which adapts to time-varying arrival rates without prior traffic knowledge. The proposed algorithm performs online estimation of request intensity based on observed interarrival times. Using Markov Modulated Poisson Process (MMPP)-based synthetic workloads with diverse request types, we conduct a simulation-based evaluation demonstrating that the proposed method achieves higher throughput than Sarathi-Serve [2], ORCA [3], and vLLM [4] in the evaluated low arrival-rate shift scenarios while maintaining comparable latency.

Related papers

• FlowADMM: Plug-and-play ADMM with Flow-based Renoise-Denoise Priors [2.986668330518244]
  FlowADMM is a state-of-the-art flow-based generative algorithm for solving inverse problems.<n>We show that FlowADMM achieves a state-of-the-art performance among flow-based generative methods on a range of inverse problems.
  arXiv  Detail & Related papers  (2026-05-09T03:21:01Z)
• Direct Preference Optimization with Rating Information: Practical Algorithms and Provable Gains [67.71020482405343]
  We study how to design algorithms that can leverage additional information in the form of rating gap.<n>We present new algorithms that can achieve faster statistical rates than DPO in presence of accurate rating gap information.
  arXiv  Detail & Related papers  (2026-01-31T08:38:21Z)
• Time-Embedded Algorithm Unrolling for Computational MRI [4.511561231517167]
  We propose a time-embedded algorithm unrolling scheme for inverse problems.<n>Our method effectively reduces aliasing artifacts and mitigates noise amplification, achieving state-of-the-art performance.
  arXiv  Detail & Related papers  (2025-10-18T03:10:09Z)
• One-Step Flow Policy Mirror Descent [52.31612487608593]
  Flow Policy Mirror Descent (FPMD) is an online RL algorithm that enables 1-step sampling during flow policy inference.<n>Our approach exploits a theoretical connection between the distribution variance and the discretization error of single-step sampling in straight flow matching models.
  arXiv  Detail & Related papers  (2025-07-31T15:51:10Z)
• Can Prompt Difficulty be Online Predicted for Accelerating RL Finetuning of Reasoning Models? [65.18157595903124]
  This work investigates iterative approximate evaluation for arbitrary prompts.<n>It introduces Model Predictive Prompt Selection (MoPPS), a Bayesian risk-predictive framework.<n>MoPPS reliably predicts prompt difficulty and accelerates training with significantly reduced rollouts.
  arXiv  Detail & Related papers  (2025-07-07T03:20:52Z)
• Optimization of Iterative Blind Detection based on Expectation Maximization and Belief Propagation [29.114100423416204]
  We propose a blind symbol detection for block-fading linear inter-symbol channels. We design a joint channel estimation and detection scheme that combines the study expectation algorithm and the ubiquitous belief propagation algorithm. We show that the proposed method can learn efficient schedules that generalize well and even outperform coherent BP detection in high signal-to-noise scenarios.
  arXiv  Detail & Related papers  (2024-08-05T08:45:50Z)
• Age-Based Scheduling for Mobile Edge Computing: A Deep Reinforcement Learning Approach [58.911515417156174]
  We propose a new definition of Age of Information (AoI) and, based on the redefined AoI, we formulate an online AoI problem for MEC systems. We introduce Post-Decision States (PDSs) to exploit the partial knowledge of the system's dynamics. We also combine PDSs with deep RL to further improve the algorithm's applicability, scalability, and robustness.
  arXiv  Detail & Related papers  (2023-12-01T01:30:49Z)
• Posterior Sampling with Delayed Feedback for Reinforcement Learning with Linear Function Approximation [62.969796245827006]
  Delayed-PSVI is an optimistic value-based algorithm that explores the value function space via noise perturbation with posterior sampling. We show our algorithm achieves $widetildeO(sqrtd3H3 T + d2H2 E[tau]$ worst-case regret in the presence of unknown delays. We incorporate a gradient-based approximate sampling scheme via Langevin dynamics for Delayed-LPSVI.
  arXiv  Detail & Related papers  (2023-10-29T06:12:43Z)
• Beyond Exponentially Fast Mixing in Average-Reward Reinforcement Learning via Multi-Level Monte Carlo Actor-Critic [61.968469104271676]
  We propose an RL methodology attuned to the mixing time by employing a multi-level Monte Carlo estimator for the critic, the actor, and the average reward embedded within an actor-critic (AC) algorithm. We experimentally show that these alleviated restrictions on the technical conditions required for stability translate to superior performance in practice for RL problems with sparse rewards.
  arXiv  Detail & Related papers  (2023-01-28T04:12:56Z)
• Accurate non-stationary short-term traffic flow prediction method [0.0]
  This paper proposes a Long Short-Term Memory (LSTM) based method that can forecast short-term traffic flow precisely. The proposed method performs favorably against other state-of-the-art methods with better performance on extreme outliers, delay effects, and trend-changing responses.
  arXiv  Detail & Related papers  (2022-05-01T17:11:34Z)
• Better than the Best: Gradient-based Improper Reinforcement Learning for Network Scheduling [60.48359567964899]
  We consider the problem of scheduling in constrained queueing networks with a view to minimizing packet delay. We use a policy gradient based reinforcement learning algorithm that produces a scheduler that performs better than the available atomic policies.
  arXiv  Detail & Related papers  (2021-05-01T10:18:34Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.