arXiv subjects

Rajkumar Buyya

Publications and source records attributed to Rajkumar Buyya.

At least 19 recordsLinked to original sources

arXiv2026

Rollout Efficiency in Reinforcement Learning for Reasoning Large Language Models: A Taxonomy and Future Directions

Reasoning-oriented reinforcement learning enables large language models to solve mathematical, coding, and other multi-step tasks, but shifts a substantial portion of the training cost to rollout, where trajectories are generated for policy updates. Efficient rollout mechanisms are therefore essential to reduce this cost while maintaining the freshness, consistency, and statistical validity of training data. This survey provides a systematic taxonomy of recent research on rollout efficiency for reasoning-oriented reinforcement learning, classifying existing approaches from both mechanism and bottleneck perspectives. Based on this taxonomy, we analyze how different technique families address distinct sources of rollout inefficiency, examine opportunities and potential conflicts for combining them, identify gaps in the evaluation and reporting of efficiency gains, and discuss open challenges and future research directions.

cs.AI↗

arXiv2026

Stability-Aware Proactive Autoscaling Using a Double Deep Q-Network in Cloud Computing Environments

Dynamic workloads and latency-sensitive applications require efficient autoscaling in cloud computing environments. However, most existing approaches rely on reactive mechanisms based on static thresholds, resulting in delayed responses and scaling oscillations under workload uncertainty. To address these limitations, we propose a double deep Q-Network-based proactive autoscaling approach (DDQN-Proactive) along with Resource Removal Strategy (RRS). The proposed (DDQN+RRS) enhances decision-making by decoupling action selection from value evaluation, enabling more stable and adaptive scaling. Experimental results demonstrate that the proposed method outperforms both reactive and existing proactive approaches. Specifically, DDQN+RRS achieves a lower Service Level Agreement (SLA) violation rate (11.81%), higher CPU utilization (52.23%), improved scaling stability, fewer scaling events (2,488), and reduced pod restarts (1,246). Furthermore, the approach ensures smoother autoscaling behavior by significantly reducing oscillations over time (0-60 s). While reactive methods exhibit substantial fluctuations in pod allocation, Reactive reduces these variations, and DDQN+RRS achieves the most stable and smooth scaling, particularly during the 15-30 s, 40-45 s, and 55-60 s intervals.

cs.DC↗

arXiv2026

Agentic Autoscaling through Worker-Pool Orchestration for LLM-driven Text Classification in Cloud Computing Environments

The growing adoption of large language model (LLM)-based systems for large-scale text processing has created a critical need for dynamic autoscaling to manage high-latency, bursty, and computationally intensive workloads. This paper proposes an agentic autoscaling framework through worker-pool orchestration for LLM-driven text classification. The framework integrates a priority task queue, a dynamic pool of agent workers, a real-time metrics collector, and an application-layer autoscaler. Its classifier-agnostic design supports both zero-shot and fine-tuned language models without modifying the autoscaling logic. The framework is evaluated using Autoscaling+BART and Autoscaling+DeBERTa against static allocation and standalone RoBERTa and DistilBERT baselines. On the AG News dataset, Autoscaling+BART achieves 84.5% accuracy, while Autoscaling+DeBERTa improves it to 90.5%. On the SMS Spam Collection dataset, Autoscaling+DeBERTa achieves 99.5% accuracy, whereas Autoscaling+BART attains 84.5% accuracy with lower execution time. Overall, the proposed framework consistently outperforms the baseline approaches in resource efficiency while maintaining high classification performance, demonstrating that elastic worker-pool orchestration provides an effective and cost-efficient solution for scalable LLM-driven text classification in cloud environments.

cs.DC↗

arXiv2026

Carbon-aware Resource Management for Latency-Sensitive Cloud Computing Environments: A Taxonomy and Future Directions

Proliferation of cloud-based latency-sensitive workloads requires infrastructures tuned to their workload-specific latency constraints. Today, they shape the cloud from a generalized computing platform to diverse workload-specific cloud environments. As the demand for latency-sensitive workloads increases, cloud service providers continue to scale their infrastructure, adversely increasing the carbon footprint and challenging climate-crisis-driven net-zero emission goals. Due to performance-oriented rigid deployment patterns of latency-optimizations, reducing its carbon footprint is challenging. Therefore, efficient techniques that exploit application specific opportunities are needed in that. To this end, we present a detailed taxonomy of recent literature on carbon-aware resource management in latency-sensitive cloud computing environments. Using the taxonomy, we analyze existing works discussing their optimization aspects, identify the gaps, and highlight future research directions.

cs.DC↗

arXiv2026

A Technique for Load Shifting Low-latency Applications in Multi-Region Renewables Harvesting via SMT Core Pooling

Load shifting across geographic regions to chase intermittent renewable energy availability is commonly used in reducing cloud infrastructure carbon footprint. However, it often omits low-latency applications due to high latency variances of wide area networks (WAN) that interconnect regions. This paper addresses accommodating low-latency applications into load shifting by minimizing their shifting across the WAN. We propose a technique using a hardware-software co-design approach. At the hardware level, we conduct server load matching over renewables supply peaks and valleys by deep idling physical cores in two otherwise identical server pools, with one enabling simultaneous multi-threading (SMT) in CPUs. In return, we achieve a static set of logical cores amidst energy supply dynamics, reducing the probability of workload shifting. At the software level, we efficiently chase the static set of cores for low-latency applications within regions while prioritizing best-effort applications to accommodate shifting requirements across WANs. Our approach exploits the lower performance compromise of SMT cores due to their hardware multi-threading. We implement the proposed technique with OpenStack and CPU idle states and evaluate its performance on a real experimental testbed with Azure VM arrival traces. Results show an 80% reduction in offloading low-latency VMs and a 43.81% reduction in coefficient of variation of p90 end-user latency while having a worst-case latency compromise of 11.97% due to SMT cores.

cs.DC↗

arXiv2026

CLASP: Chained-Request-Aware Scaling and Operator Placement for Serverless Stream Processing

Stateful serverless (Function-as-a-Service) environments, whose workers host state servers, are increasingly used for stream processing. A stream application is a pipeline of operators, where each operator forwards intermediate data downstream through a chained request. As input rates fluctuate, the system should adjust operator parallelism and place instances across workers to sustain the incoming rate. Existing approaches do so without fully accounting for chained-request overhead, leading them to misestimate the required number of workers. Too few leave the cluster unable to keep up with the input rate, while too many route a larger fraction of chained requests across worker boundaries, increasing end-to-end latency. We propose CLASP, a scaling and scheduling strategy for stream processing in stateful serverless environments. At runtime, CLASP estimates execution cost and chained-request cost from observed metrics. Under a capacity model that covers the two costs, it adjusts operator parallelism and packs operators onto the fewest workers that can sustain the target input rate. Once a scaling decision is made, CLASP migrates each operator's state together with its instances, thereby minimizing execution pause time. Experiments show that CLASP improves throughput by up to 3.3x and reduces median end-to-end latency by up to 76% compared with state-of-the-art scaling strategies.

cs.DC↗

arXiv2026

SketchGuard: Scaling Byzantine-Robust Decentralized Federated Learning via Sketch-Based Screening

Byzantine-robust decentralized federated learning (DFL) protects peer-to-peer training from malicious clients. The dominant defenses rely on similarity-based filtering, in which each client exchanges full model vectors with every neighbor before any filtering decision; this communication grows with the model dimension and scales poorly as models grow. We propose SketchGuard, which decouples screening from aggregation: clients screen neighbors in a compact Count Sketch domain and fetch full models only from those that pass the screen. We show this idea is insecure when implemented naively. Because the sketch is a fixed, publicly known linear map, an adaptive adversary can hide an arbitrarily large perturbation in its null space, so the poisoned model passes both the sketch-domain filter and the re-sketch verification. We prove this vulnerability and close it with commit-then-sketch, a one-message protocol that draws the sketch seed only after models are committed, restoring the oblivious setting in which Count Sketch provably preserves screening decisions. We then establish convergence in strongly convex and non-convex settings, with explicit dependence on network connectivity and data heterogeneity. Empirically, secured SketchGuard matches state-of-the-art full-precision robustness, up to a small threshold inflation, across six attacks including the adaptive null-space attack, a range of network topologies and heterogeneity settings, and a decentralized fine-tuning task on an 11-million-parameter language model, while reducing per-neighbor screening communication to a size independent of the model dimension.

cs.LG↗

arXiv2026

Topology-Aware Differential Privacy in Hierarchical Federated Learning

Hierarchical federated learning places regional aggregators between clients and the cloud, so a participant's update is observed only alongside its neighbours'. The concealment this arrangement provides depends on the size of the aggregation region, and regions in operational deployments vary widely. Prevailing practice applies a single noise multiplier to every participant, calibrated for the most exposed region, so every other participant carries more noise than its own exposure requires. We show that this allocation problem admits an explicit solution. We first give a silo-level differential privacy guarantee for the mechanism, then bound the mutual information between a participant's local class distribution and any estimate an observer positioned above the regional tier could form of it, using an adjacency notion matched to the quantity being protected. Minimising the worst-case bound under a fixed utility budget yields a min-max optimal allocation, which we call Fulcrum. The budget it recovers has a closed form we term the exposure dispersion, a measure of how unevenly aggregation weight is concentrated within regions relative to the most exposed one. Because this quantity follows from the region structure and the aggregation weights alone, a practitioner can evaluate it before training begins, and it vanishes precisely when all regions are equally exposed. On image and text classification at $\varepsilon = 0.99$, accuracy at a matched worst-case per-client guarantee improves by up to $14.84$ and $12.16$ percentage points where the dispersion is large, and is exactly zero on a balanced control for which the theory predicts parity.

cs.CR↗

arXiv2026

AgentR A Stateful and Recovery-Aware Software Architecture for LLM-based Auditable Workflows

Modern LLM-based applications increasingly require multi- stage execution, persistent intermediate state, retry seman- tics, and auditable usage accounting. However, many LLM applications are still implemented as stateless prompt- response wrappers or session-bounded conversational sys- tems, which makes them difficult to recover, audit, and re- produce after interruption or failure. We propose AgentR, a stateful architecture for LLM workflow systems that en- ables persistence and recovery, instantiated through scien- tific literature review as a representative use case. AgentR represents research intent, generated queries, candidate- paper assessments and gap analyses as durable workflow artifacts, and executes the pipeline through asynchronous BullMQ workers backed by Redis, with PostgreSQL as the persistence store. The design includes explicit processing state transitions, retries with exponential backoff, orphan job detection, credit-aware pre-checks, ACID token-cost logging, and Type-2 slowly changing pricing records. We evaluate AgentR on telemetry collected from a prototype deployment. At the LLM stage, the system achieves 99.2% job completion, and mean latencies of 9.0 s, 18.9 s, and 25.4 s for intent decomposition, query generation, and paper scoring, respectively. Parallel scoring allows for analytical latency modeling from observed calls, leading to as much as 4.3 wall-clock speedup over sequential execution. The results provide preliminary proof-of-concept that persistent state machine design, asynchronous orchestration, and cost-aware usage logging can enable improved observability, recoverability, and operational accountability in LLM workflow systems. The prototype implementation of AgentR is publicly available at: https://github.com/ RiyaSamanta/AgentR-public.

cs.SE↗

arXiv2026

How Much Reconstruction Does Quantum Machine Learning Need? Late Fusion of Independently Trained Quantum Subcircuits

Circuit cutting lets a large quantum neural network (QNN) run as independent subcircuits on small devices, but rebuilding its outputs by reconstruction carries a classical sampling overhead exponential in the number of cuts - the dominant runtime cost in prior work. We ask whether, for machine-learning tasks, this step is necessary, and replace it with late fusion: each subcircuit is trained and measured independently, and a small classical head combines their outputs - a linear-cost, decision-level combination borrowed from multimodal learning. To characterize the trade-off we introduce a quantumness dial $Q$, a tunable reconstruction budget interpolating from pure fusion to full reconstruction, and a cut-entanglement diagnostic that indicates how much reconstruction a task needs (Spearman $ρ=0.59$ over $104$ runs). Across synthetic and standard datasets, independently trained late fusion matches full reconstruction accuracy within $0.04$ at every point of the controlled sweep and on every classical benchmark, at exponentially lower cost; it is also markedly more robust to shot and device noise. Controlled entangled-data experiments locate the boundary where fusion must fail. We do not claim advantage over classical machine learning - consistent with recent benchmarking, quantum offers no accuracy edge on these datasets. Late fusion is thus an efficient, noise-robust, self-characterizing alternative to reconstruction for circuit-cutting QML.

quant-ph↗

arXiv2026

Trust-Aware Topology Learning for Dynamic Decentralized Federated Learning under Adversaries

In dynamic mobile decentralized federated learning (DFL), adversaries can poison both model updates and the topology information devices use to choose collaborators. We present DMTT (Dynamic MURMURA with Trusted Topology), a decentralized personalized FL protocol built on MURMURA, which uses evidential deep learning to down-weight distribution-mismatched peers, extended here to time-varying graphs under topology-manipulation attacks. Each device maintains a confidence-weighted local topology view from link-reliability estimates, signed topology claims, witness corroboration, and a Beta-distributed source-trust model, then aggregates only over a trust-screened collaborator set using a composite score fusing model compatibility, topology trust, and link reliability. We prove the screened mixing matrices confine Byzantine influence to a bounded residual $δ_{max}$ that vanishes under perfect screening, and implement DMTT as a coordinator-free distributed system with each client running as an independent ZeroMQ process synchronized by a shared wall-clock epoch. On UCI HAR and PAMAP2, each partitioned across 100 mobile clients with Dirichlet heterogeneity, DMTT sustains honest-node accuracy above 0.862 (UCI HAR) and 0.829 (PAMAP2) across all tested adversary fractions (10 to 80%), nearly matching no-attack accuracy at low fractions and degrading gracefully toward local-only performance at extremes; static and dynamic FedAvg collapse to chance at every fraction, and robust aggregators (Krum, BALANCE, UBAR) fail to consistently beat a local-only baseline, while DMTT is the only method that clears this bar across both datasets at all fractions, with surviving Byzantine aggregation weight empirically zero throughout, consistent with $δ_{max}$=0. The protocol runs end-to-end on real nodes via a coordinator-free ZeroMQ backend on the Melbourne Research Cloud.

cs.DC↗

arXiv2026

FedCARE: A Multi-Objective Personalised Federated Learning Framework for Smart Healthcare

Federated Learning (FL) enables collaborative model training across distributed healthcare institutions without centralising sensitive patient data. However, real-world healthcare federations are often characterised not only by non-IID data, but also by heterogeneous clinical objectives and partially overlapping feature spaces. Different hospitals may optimise distinct and potentially conflicting objectives, such as mortality risk prediction, readmission reduction, or length-of-stay estimation, while also retaining institution-specific clinical features that cannot be shared with other participants. Existing personalised FL methods mainly address statistical heterogeneity, whereas multi-objective FL approaches typically learn a shared global model without explicit client-level adaptation. To address these limitations, we propose \textbf{FedCARE}, a multi-objective personalised FL framework for smart healthcare services. FedCARE follows a two-stage training strategy. First, it learns a shared global backbone from common clinical features using Pareto-driven multi-objective federated optimisation. Second, each client independently fine-tunes the shared backbone using its private features and local clinical objectives, enabling institution-specific personalisation without additional communication overhead. We implement FedCARE in a cloud-based client-server federated deployment on the Melbourne Research Cloud and evaluate it on two real-world healthcare datasets, MIMIC-III and Diabetes 130-US Hospitals. Experimental results show that FedCARE consistently outperforms standard FL, multi-objective FL, and personalised FL baselines, achieving up to 12.5% AUROC improvement and 32.0% MAE reduction over FedAvg.

cs.LG↗

arXiv2026

PrefixPlace: Provable Prefix Key-Value Placement for Large Language Model Serving under Heterogeneous Compute and Transfer Costs

Prefix Key-Value (KV) reuse avoids repeated prefill in Large Language Model (LLM) inference, but local misses require recomputation or replica fetches. Their relative cost varies with hardware, prefix depth, KV goodput, and replica location, making hit-rate-based placement suboptimal. To address this issue, we propose an epoch-level planner, PrefixPlace, which assigns prefix-complete targets under memory budgets and profiled demand, compute, and transfer costs. The objective decomposes into local-copy value plus first-replica coverage, and source-dependent costs yield a monotone facility-location objective; each worker update is an additive rooted-tree problem solved exactly in O(nk) time for n chunks and capacity k, giving a fixed-order 1/2-approximation that coordinate refinement and order-diverse starts improve without weakening. T4, L4, and A100 measurements reveal distinct regimes. Across 432 instances with exact optima, PrefixPlace averages 99.84% of optimum and never falls below 98.02%. In Retrieval-Augmented Generation (RAG) replays, it improves materialization-cost saving by 40.3% over vLLM Automatic Prefix Caching (vLLM-APC) and 6.3% over the best offline baseline. On WikiQA, gains are 40.4% and 5.3%. Finally, PrefixPlace solves a 50,000-node, 16-worker placement in 12.3 s on one processor, enabling timely replanning.

cs.DC↗

arXiv2026

Preserving Admission Responsibility in Multi-Tenant Large Language Model Prefix Caches

Shared prefix caching turns Graphics Processing Unit (GPU) memory into persistent state shared across Large Language Model (LLM) tenants. A group that materializes new Key-Value (KV) blocks can force another to lose reusable state, yet request-time schedulers account for transient service, replacement policies primarily rank object value, and static partitioning strands idle capacity. We call this mismatch the admission-responsibility gap. To close it, we propose PrefixShield, which meters newly materialized full KV blocks, carries responsibility across requests, gates reuse promotion while debt remains, and uses projected debt to select the group supplying eviction candidates. We implement PrefixShield in vLLM. In paired runs under one-touch pollution, PrefixShield improves victim cache hit ratio by 9.39 percentage points over the Least Recently Used (LRU) policy and 8.64 points over S3-FIFO, restoring the victim from 4.92% to 84.87% at 4096-block scale, and gains 2.00 points over S3-FIFO under two-pass replay. It preserves benign ShareGPT behavior and work-conserving access to idle capacity. Delayed replay yields a 35.16-point advantage while debt remains. These results show that object-value signals rank what to retain, while persistent responsibility determines which group bears reclamation pressure.

cs.DC↗

arXiv2026

ADASCALE: An Adaptive Scaling and Placement Framework for Microservices Under Dynamics

Microservice applications are increasingly deployed across cloud--edge environments, where heterogeneous nodes and time-varying inter-node delays amplify the impact of placement decisions. At the same time, these applications face non-stationary traffic, shifts in the mix of root request operations that exercise different call graphs, and heterogeneous communication modes that determine how network latency and queuing propagate to end-to-end (E2E) performance. Existing autoscalers and network-aware schedulers typically handle only a subset of these dynamics, leading to either compute bottlenecks or inflated cross-node latency and thus SLO violations. We propose ADASCALE, an adaptive framework that jointly scales and places microservice replicas under such multi-dimensional dynamics. ADASCALE implements a Monitor--Analyzer--Planner--Executor (MAPE) loop that extracts per-edge and per-service demand from distributed traces and service-mesh metrics, identifies the most critical root operation under a mixed workload, computes SLO-aware replica targets, and then places replicas to minimize a demand-weighted latency objective given the current inter-node latency matrix. To react quickly to networking perturbations, ADASCALE triggers a reactive placement loop, while a steady-state autoscaling loop handles demand shifts. We evaluate ADASCALE on a cloud--edge Kubernetes cluster using the DeathStarBench Social Network application with three root operations under varying load and workload mixes. Across scenarios, ADASCALE consistently meets SLO targets and improves both latency and throughput: compared with NetMARKS_Scale, it achieves up to 1.56x, 1.93x, and 1.34x lower average response time (for compose-post, read-home-timeline, and read-user-timeline) and up to 2.16x, 1.32x, and 1.36x higher throughput, respectively.

cs.NI↗

arXiv2026

ReinFog: A Deep Reinforcement Learning Empowered Framework for Resource Management in Edge and Cloud Computing Environments

The growing IoT landscape requires effective server deployment strategies to meet demands including real-time processing and energy efficiency. This is complicated by heterogeneous, dynamic applications and servers. To address these challenges, we propose ReinFog, a modular distributed software empowered with Deep Reinforcement Learning (DRL) for adaptive resource management across edge/fog and cloud environments. ReinFog enables the practical development/deployment of various centralized and distributed DRL techniques for resource management in edge/fog and cloud computing environments. It also supports integrating native and library-based DRL techniques for diverse IoT application scheduling objectives. Additionally, ReinFog allows for customizing deployment configurations for different DRL techniques, including the number and placement of DRL Learners and DRL Workers in large-scale distributed systems. Besides, we propose a novel Memetic Algorithm for DRL Component (e.g., DRL Learners and DRL Workers) Placement in ReinFog named MADCP, which combines the strengths of Genetic Algorithm, Firefly Algorithm, and Particle Swarm Optimization. Experiments reveal that the DRL mechanisms developed within ReinFog have significantly enhanced both centralized and distributed DRL techniques implementation. These advancements have resulted in notable improvements in IoT application performance, reducing response time by 45%, energy consumption by 39%, and weighted cost by 37%, while maintaining minimal scheduling overhead. Additionally, ReinFog exhibits remarkable scalability, with a rise in DRL Workers from 1 to 30 causing only a 0.3-second increase in startup time and around 2 MB more RAM per Worker. The proposed MADCP for DRL component placement further accelerates the convergence rate of DRL techniques by up to 38%.

cs.DC↗

arXiv2026

Security and Privacy in Retrieval-Augmented Generation: Architectures, Threats, Defenses, and Future Directions for Building Trustworthy Systems

Retrieval-Augmented Generation (RAG) has emerged as a dominant paradigm for enhancing large language models with external knowledge. By coupling retrieval mechanisms with generative models, RAG systems improve factual grounding and adaptability across domains. However, integrating retrieval pipelines introduces new security and privacy risks that extend beyond conventional language modeling threats. Sensitive information may be exposed through retrieval indices, query logs, context construction, or federated updates, while adversarial manipulation of knowledge bases can undermine trust in generated outputs. This survey provides a comprehensive examination of privacy and security challenges across RAG systems deployed in centralized, on-device (Micro-RAG), federated, and hybrid paradigms. We present a unified taxonomy of threat surfaces spanning the retrieval, context construction, and generation stages and systematically analyze attack classes, including membership inference, index inference, poisoning, gradient leakage, and collusion. We further review architectural, algorithmic, and cryptographic defenses, highlighting privacy-utility trade-offs and deployment considerations. Finally, we outline open research challenges toward building trustworthy, secure, and resilient RAG systems for real-world applications.

cs.CR↗

arXiv2026

iDynamics: A Configurable Emulation Framework for Evaluating Microservice Scheduling Policies under Controllable Cloud-Edge Dynamics

This paper presents iDynamics, a configurable emulation framework that exposes these dynamics as controllable experimental factors while running real microservice code on a Kubernetes-based cloud-edge cluster. iDynamics comprises three modular components. The Graph Dynamics Analyzer reconstructs application call graphs from service-mesh telemetry and quantifies bidirectional traffic between upstream-downstream microservice pairs. The Networking Dynamics Manager injects and measures realistic cross-node delay and bandwidth patterns via Linux traffic control primitives and distributed agents. The Scheduling Policy Extender offers a pluggable interface and utility library for implementing and evaluating arbitrary scheduling policies, expressed as pod placement and migration strategies. We use iDynamics to implement two representative policies -- a call-graph-aware policy and a hybrid policy that jointly considers traffic and latency -- as case studies demonstrating how the framework can be used to study SLA compliance under dynamic conditions. Experiments on a real cloud-edge cluster, running the DeathStarBench Social Network microservices, show that iDynamics can accurately emulate targeted network conditions, generate diverse call-graph and traffic patterns, and help quantify how different scheduling policies mitigate SLA violations under controllable and repeatable dynamics.

cs.DC↗