A research team at Changsha University of Science and Technology in China has now tackled this question with an unusually comprehensive strategy. In a study published in Cluster Computing, Qiang Tang, Defa Ouyang, Heng Liu, and Lei Yang present a framework called VDDM, short for VM Dynamic partition and DRL-based Delay Minimization. The work targets a scenario with multiple drones serving multiple user equipments, the phones, sensors, and devices that generate computational workloads. The goal is deceptively simple to state and notoriously hard to achieve: minimize the total task delay experienced by all users simultaneously, while juggling virtual machine configuration, user-to-drone assignments, and the flight paths of the aircraft themselves.

The mathematical backbone of the study is a mixed-integer nonlinear programming problem, a class of optimization that combines continuous variables, such as a drone’s position in three-dimensional space, with discrete choices, such as which user connects to which drone and how many virtual machines each drone should run. Problems of this type are computationally brutal; the search space explodes combinatorially as the number of drones, users, and time slots grows. Solving them exactly is often infeasible in real time, which is precisely why the authors decomposed the challenge into three cooperating components rather than attacking it as one monolithic calculation.

The first component addresses a subtle but performance-critical issue that has plagued virtualized systems for years: input-output interference among virtual machines. When several VMs on the same physical host hammer the disk and network subsystems at once, they contend for shared I/O channels, and each one slows the others down. Prior research, including widely cited work on performance interference in virtualized cloud environments, has documented how badly I/O-heavy workloads can degrade when packed onto a single host. The VDDM framework counters this with a dynamic partition algorithm that adjusts the number of virtual machines on each drone according to the sizes of the tasks being processed. Instead of a fixed VM configuration, the partition adapts to the workload, reducing interference and keeping the drone’s parallel computing capacity genuinely usable.

The second component handles the matching problem between users and drones. In a multi-UAV network, every user wants to offload to the drone that will serve it fastest, but if everyone piles onto the same aircraft, that drone becomes a bottleneck while its neighbors idle. The researchers apply matching theory, a mathematical framework originally developed for economics problems like stable marriages and school assignments, to pair each user equipment with the UAV carrying the lowest load. This load-aware assignment spreads the computational demand across the fleet, preventing hotspots and smoothing out the queueing delays that build up when too many tasks land on one server at once.

The third and perhaps most visually striking component is the flight control layer. Where a drone hovers determines the quality of its wireless channel to each user, and therefore how fast data can flow during offloading. Optimizing trajectories jointly with everything else is what elevates this work beyond simple resource scheduling. The authors turn to deep reinforcement learning, specifically the Soft Actor-Critic algorithm, a technique that has become a favorite in control problems because it balances exploration of new actions against exploitation of known good ones. SAC learns a policy that guides each UAV’s flight path over time, continuously repositioning the aircraft to shorten transmission distances, improve channel conditions, and ultimately cut the end-to-end delay of every offloaded task.

Deep reinforcement learning deserves a moment of explanation for readers unfamiliar with it. In this paradigm, an agent, here the drone’s onboard decision system, observes the state of its environment, takes an action such as moving to a new coordinate, and receives a reward signal that reflects how well that action served the overall objective of minimizing delay. Over many training episodes, the neural network behind the agent learns which maneuvers and configurations yield the best long-term outcomes, even in environments too complex and dynamic for hand-written rules. SAC adds entropy to the reward, encouraging the agent to keep trying diverse actions rather than collapsing prematurely onto a mediocre strategy, which is one reason the approach tends to converge reliably in high-dimensional problems like multi-UAV trajectory planning.

The three components of VDDM do not operate in isolation; they form a pipeline in which each stage feeds the next. Task sizes inform the virtual machine partition, the partition shapes each drone’s effective capacity, capacity determines the sensible user-to-drone matches, and the resulting pattern of communication demands shapes the trajectories the SAC agent learns to fly. By formulating the whole system as a single MINLP objective before decomposing it, the authors ensure that the pieces pull in the same direction rather than optimizing themselves into conflict. This joint design philosophy reflects a broader trend in edge computing research, where treating offloading, resource allocation, and mobility as separate problems has repeatedly been shown to leave performance on the table.

How well does it actually work? According to the team’s extensive simulation results, VDDM significantly outperforms existing benchmark strategies in the two metrics that matter most for this class of system: total task delay and training convergence speed. The delay reduction follows directly from the framework’s ability to simultaneously cut I/O interference, balance loads across the fleet, and physically move computing power closer to the users who need it most. The faster convergence is a practical bonus with real consequences, because reinforcement learning systems that train slowly burn time and energy before they become useful, and in dynamic environments a policy that converges quickly adapts to changing user distributions sooner. The study reports its findings across a series of figures and algorithm descriptions that trace how each component contributes to the overall gain.

The implications stretch well beyond the simulation lab. Multi-UAV edge computing is a candidate technology for emergency response networks, where terrestrial base stations may be destroyed; for temporary mass gatherings, where cellular capacity buckles under demand; for precision agriculture and infrastructure inspection across vast rural areas; and for maritime and border surveillance, where earlier work by some of the same authors has explored latency minimization in aerial networks. In each of these settings, delay is not merely a quality-of-service statistic but often the difference between a useful system and a useless one. A framework that squeezes latency out of every layer of the stack, from the virtual machine scheduler to the flight controller, moves the field measurably closer to deployments that can be trusted under pressure.

The work also highlights a challenge the field is only beginning to confront honestly: the gap between idealized virtual machines and the messy physical reality of shared hardware. Much of the existing literature on UAV-assisted MEC treats a drone’s computing capacity as a clean, divisible quantity, ignoring the contention effects that arise when multiple workloads share the same I/O subsystem. By building I/O interference into the problem formulation and designing the dynamic partition algorithm specifically to mitigate it, the Changsha team grounds its optimization in a more realistic model of what actually happens inside a virtualized host. That realism, combined with learning-based trajectory control, suggests a template for future systems in which flying edge servers are not just a research curiosity but dependable infrastructure, quietly hovering overhead and making sure the computations that matter arrive on time.

Subject of Research: Delay minimization for computation offloading in multi-UAV assisted mobile edge computing using virtual machine partitioning and deep reinforcement learning

Article Title: A VM dynamic partition and drl based delay minimization offloading strategy for Multi-UAV assisted MEC

Article References: Tang, Q., Ouyang, D., Liu, H., & Yang, L. (2026). A VM dynamic partition and drl based delay minimization offloading strategy for Multi-UAV assisted MEC. Cluster Computing, 29(13), Article 737. https://doi.org/10.1007/s10586-026-06512-z

Image Credits: AI Generated

DOI: 10.1007/s10586-026-06512-z

Keywords: mobile edge computing, unmanned aerial vehicles, virtual machines, deep reinforcement learning, Soft Actor-Critic, trajectory optimization, computation offloading, I/O interference, matching theory, mixed-integer nonlinear programming, task delay, UAV networks

In a paper published in Cluster Computing, Jasmine Kaur, Inderveer Chana and Anju Bala introduce an enhanced version of a serverless simulation framework that adds a virtual machine layer between physical servers and the containers that actually run user code. On top of that more realistic architecture, they bolt on a multi-agent reinforcement learning technique known as proximal policy optimization, or PPO, to decide where workloads should go. The results are striking: cold start latency drops by up to 29 percent, the number of active physical machines falls by as much as 66 percent, and energy consumption shrinks by up to 23 percent. For an industry under mounting pressure to curb the enormous electricity appetite of data centers, those numbers carry real weight.

To understand why the virtual machine layer matters, it helps to look at how serverless platforms are actually built. In principle, containers, lightweight packages of software that share the host operating system’s kernel, could run directly on physical machines. That is precisely how ServlessSimPro, the earlier simulation platform on which the new work builds, modeled the world. But in production deployments at major cloud providers and in popular open-source platforms such as Knative and OpenFaaS, containers are almost always nested inside virtual machines. The VM adds an extra boundary of isolation and security, but it also introduces an additional layer of resource management, with its own scheduling decisions, consolidation opportunities and performance overheads.

By ignoring that layer, earlier simulators produced results that diverged from what operators actually observe. A scheduling algorithm that looks brilliant in a flat, container-on-metal simulation may behave very differently when it must first choose a virtual machine, and only then a container within it. The enhanced framework bridges this architectural gap, giving researchers a testbed whose resource abstractions mirror the deployment models used in the real world. That matters because simulation is where most scheduling research lives: few academic groups can run experiments on hyperscale infrastructure, so the fidelity of the simulator effectively determines how transferable their ideas are.

The second half of the contribution is the scheduling brain itself. Rather than a single centralized controller making every placement decision, the researchers deploy multiple reinforcement learning agents, each powered by proximal policy optimization, working in parallel across the simulated cluster. PPO is a policy-gradient method that has become a favorite of the reinforcement learning community because it makes steady, clipped updates to its decision policy, avoiding the catastrophic performance collapses that can plague more aggressive learning algorithms. Each agent is responsible for a slice of the scheduling problem: deciding where containers should be placed, when virtual machines should be consolidated onto fewer physical hosts, and how incoming workloads should be distributed across the fleet.

Decentralization is not just an aesthetic choice. In a large data center, a single scheduler becomes a bottleneck and a single point of failure, and the state of the system grows so large that no one agent can see everything at once. Multi-agent architectures let decisions happen closer to the resources they affect, and the multi-agent formulation of PPO, often abbreviated MAPPO, allows the agents to learn coordinated behavior despite their partial views. The approach builds on the authors’ earlier work on multi-agent deep Q-learning for serverless job scheduling, but PPO’s more stable training dynamics make it better suited to the high-dimensional, continuous decisions involved in juggling containers, VMs and workloads simultaneously.

The headline metric, cold start latency, deserves particular attention because it is the Achilles’ heel of serverless computing. When a function is invoked and no warm instance exists, the platform must provision a container from scratch, an operation that can add hundreds of milliseconds or more to response time. For latency-sensitive applications, from interactive web services to real-time data pipelines, those milliseconds are the difference between a seamless experience and a visibly sluggish one. By learning placement and consolidation policies that keep likely-to-be-needed functions warm and co-located with available capacity, the MAPPO-driven scheduler cuts cold starts by nearly a third in the reported experiments.

The efficiency gains are equally significant. Consolidating workloads so that up to 66 percent fewer physical machines need to stay active is a dramatic reduction, and it cascades directly into the 23 percent cut in energy consumption. Data centers are among the fastest-growing consumers of electricity worldwide, and idle servers draw substantial power even when they process nothing. Any technique that lets operators switch off more of the fleet, without degrading the responsiveness users feel, translates into both lower costs and lower carbon emissions. The fact that the framework achieves these savings while simultaneously improving latency suggests the agents are finding genuinely smarter trade-offs, not merely shifting costs from one column to another.

The study also situates itself within a rapidly maturing ecosystem of simulation tools. Open-source options such as FaasSim, FaaS-Sim and SimFaaS have given researchers starting points for modeling function-as-a-service platforms, and commercial offerings from AWS, Microsoft Azure, Google Cloud and IBM have defined the de facto behavior the simulators try to capture. What has been scarce, the authors argue, is a comprehensive platform that combines realistic multi-layer resource modeling with the ability to train and evaluate learning-based schedulers. By pairing the VM-aware architecture with MAPPO, the framework becomes both a measurement instrument and a training ground, letting researchers prototype scheduling policies that can be benchmarked consistently before any real-world trial.

The broader significance of the work lies in the convergence of two trends: the industrial consolidation around serverless architectures and the rapid advance of multi-agent reinforcement learning as a practical tool for systems management. As edge computing, Kubernetes-based orchestration and energy-aware provisioning continue to reshape how cloud infrastructure is organized, the ability to simulate those systems faithfully, and to let learning agents discover scheduling policies that human engineers might never hand-design, could define the next generation of cloud efficiency. The Thapar Institute team’s framework offers the research community a way to explore that future at scale, on a laptop rather than a data center, and with an architectural honesty that earlier tools lacked. If the reported gains hold when such policies migrate from simulation to production clusters, the invisible machinery of the cloud may soon be run by algorithms that learned their craft in a simulator, one carefully modeled virtual machine at a time.

Subject of Research: VM-aware serverless simulation and multi-agent reinforcement learning for cloud job scheduling

Article Title: VM-aware serverless simulation framework with multi-agent proximal policy optimization for scalable and efficient job scheduling

Article References: Kaur, J., Chana, I., & Bala, A. (2026). VM-aware serverless simulation framework with multi-agent proximal policy optimization for scalable and efficient job scheduling. Cluster Computing, 29(13), Article 758. https://doi.org/10.1007/s10586-026-06585-w

Image Credits: AI Generated

DOI: 10.1007/s10586-026-06585-w

Keywords: serverless computing, simulation framework, virtual machines, multi-agent reinforcement learning, proximal policy optimization, job scheduling, cold start latency, energy efficiency, cloud computing, resource utilization, container orchestration, data centers

The algorithm, described in the journal Cluster Computing, is called Opposition-Based Firebug Tunicate Optimization, or OFTO. It is the brainchild of Chirag Chandrashekar, Pradeep Krishnadoss, Arun Kumar Sivaraman, Vijayakumar Kedalu Poornachary, Kong Fah Tee, Janakiraman Nithiyanantham and colleagues, and it addresses a problem that has dogged cloud schedulers for years: the delicate balance between exploring the search space and exploiting promising solutions once they are found. Metaheuristic algorithms — the family of nature-inspired techniques that includes particle swarm optimization, ant colony methods, and grey wolf optimizers — are widely used for scheduling because the underlying problem is combinatorial and grows explosively with the number of tasks and machines. But many of these algorithms converge prematurely, locking onto a mediocre schedule before they have surveyed the full landscape of possibilities.

OFTO tackles that weakness by fusing two existing swarm intelligence techniques, each with complementary strengths. The first is Firebug Swarm Optimization, an algorithm modeled on the mating behavior of firebugs, which excels at local search — refining a candidate solution in its immediate neighborhood. The second is Tunicate Swarm Optimization, inspired by the jet-propulsion movement and swarming of marine tunicates, which provides robust global search, allowing the population to leap across distant regions of the solution space. By combining the two, the researchers created a hybrid whose population can both probe widely and drill deeply, avoiding the twin failure modes of wandering forever without converging and of settling too early on a local optimum.

The third ingredient is what gives the algorithm its name’s prefix: opposition-based learning. The idea, first proposed in the mid-2000s, is deceptively simple. When searching for a good solution, it often pays to evaluate not just a candidate but also its mathematical opposite — a point reflected across the center of the search space. If the opposite point happens to be better, the algorithm jumps there; if not, the comparison still enriches the population’s diversity. In OFTO, every iteration evaluates both current and opposite candidate solutions, which the authors say ensures comprehensive coverage of the search space while maintaining exploitation efficiency, accelerating convergence and helping the algorithm escape local optima that would otherwise trap less adaptable schedulers.

Crucially, OFTO is not just a generic optimizer repurposed for the cloud. The researchers designed it as what they call a Deadline Sensitive Task Scheduler, tuned to the quality-of-service parameters that matter most in commercial cloud environments: execution cost, makespan — the total time to finish all tasks — and the speed at which a good schedule is found. In practice, that means the algorithm assigns each incoming task to the virtual machine best positioned to complete it within a constrained time frame while consuming minimal computational resources. For cloud providers, whose profit margins depend on keeping utilization high and service-level agreements intact, that combination of speed, cost, and deadline awareness is precisely the trade-off they need to manage.

To test their scheduler, the team ran extensive simulations against a lineup of state-of-the-art competitors, including a fault-tolerant trust-based Harris Hawks deep reinforcement learning algorithm known as FTTHDRLA, a modified transfer-function-based binary particle swarm optimizer called MTF-BPSO, the Rider Cuckoo Optimization Algorithm, and the two parent algorithms, Tunicate Swarm Optimization and Firebug Swarm Optimization, on their own. The results were striking. OFTO improved overall performance by 7.96 percent over FTTHDRLA, 14.98 percent over MTF-BPSO, 18.50 percent over RCOA, 24.75 percent over TSO, and 31.71 percent over FSO. The fact that it beat both of its own parent algorithms suggests the hybridization and the opposition mechanism each contribute something the individual components lack.

The authors did not rely on raw performance numbers alone. They subjected their results to a battery of statistical significance tests — the T-test, the Wilcoxon signed-rank test, and the Friedman test — which are standard tools for establishing that one optimizer’s advantage over another is not a fluke of a particular random seed or workload. The tests, they report, confirm the robustness and reliability of the proposed algorithm across multiple metrics, lending the headline improvements a firmer footing than a single favorable benchmark would provide.

The experimental setup itself reflects the practical constraints of cloud research. Because cloud service providers enforce strict data confidentiality policies, real user workload data is notoriously difficult for researchers to obtain, and the study was conducted without external funding, putting real-time cloud execution beyond the project’s scope. Instead, the team used an open-source benchmark, the GoCJ Google Cloud Jobs Dataset hosted on the Mendeley Data repository, which is derived from real-world Google cluster traces and MapReduce logs. That gives the simulations a credible anchor in genuine production workloads, even though the experiments were run in simulation rather than on a live commercial cloud — a caveat that readers should keep in mind when weighing the reported gains.

The work arrives amid a flourishing of hybrid metaheuristics for cloud scheduling. Recent years have seen krill herd algorithms for fog-cloud workflow scheduling, chameleon and remora search for handling scheduling uncertainty, improved wild horse optimization with Levy flight, fuzzy quantum atom search optimizers, beluga whale optimization with ring topologies, and adaptive chaotic reverse learning-enhanced reptile search, among many others. The proliferation reflects both the difficulty of the scheduling problem and the hunger of the cloud industry for any edge in efficiency. Against that crowded backdrop, OFTO’s distinguishing features are its explicit deadline sensitivity and its systematic use of opposition-based learning to keep population diversity high throughout the search.

For the researchers, the next steps will likely involve testing the algorithm under more dynamic conditions — fluctuating workloads, machine failures, and heterogeneous resource types — and, eventually, deployment on real infrastructure. For the rest of us, the study is a reminder that some of the most consequential computation in the modern world happens before our tasks even begin to run. The difference between a cloud data center that hums and one that thrashes comes down to algorithms like these: silent, insect-inspired matchmakers deciding, millions of times a second, where the world’s digital work should go. If OFTO’s reported gains hold up outside the simulator, the firebug and the tunicate may have just earned a small but real place in the plumbing of the internet.

Subject of Research: Hybrid opposition-based swarm intelligence for deadline-sensitive task scheduling in cloud computing

Article Title: Efficient task scheduling in cloud computing environment using opposition-based firebug tunicate optimization

Article References: Chandrashekar, C., Krishnadoss, P., Sivaraman, A. K., Poornachary, V. K., Tee, K. F., & Nithiyanantham, J. (2026). Efficient task scheduling in cloud computing environment using opposition-based firebug tunicate optimization. Cluster Computing, 29(14), Article 819. https://doi.org/10.1007/s10586-026-06561-4

Image Credits: AI Generated

DOI: 10.1007/s10586-026-06561-4

Keywords: cloud computing, task scheduling, swarm intelligence, metaheuristics, opposition-based learning, makespan, quality of service, virtual machines, Firebug Swarm Optimization, Tunicate Swarm Optimization, deadline scheduling, Cluster Computing

The motivation is straightforward economics and engineering. In cloud computing, host-level load prediction is described by the authors as indispensable for optimizing resource utilization, balancing load across machines, and minimizing energy consumption. Overprovision a data center and you waste electricity on idle servers; underprovision it and applications stall, latency climbs, and service-level agreements are violated. Yet forecasting host load is notoriously difficult because of two persistent obstacles the paper identifies: inefficiency in feature extraction and the sheer variability of workloads. Traffic on a cloud host is a chaotic mixture of scheduled jobs, bursty user requests, and background processes, and the signals that actually drive future load are buried inside that noise.

What distinguishes this study from many earlier forecasting efforts is its embrace of explainable artificial intelligence, or XAI. Deep learning models are famously opaque: they can produce a number, but they rarely tell an operator why. The researchers incorporated explainability techniques, most prominently SHAP, which stands for SHapley Additive exPlanations, into their framework. SHAP, originally introduced by Scott Lundberg and Su-In Lee in 2017, borrows from cooperative game theory and assigns each input feature a contribution value for a given prediction, distributing credit among features in a mathematically consistent way. In this work, the authors report that SHAP outperformed other explainability approaches due to its consistent and transparent evaluations, and they used it for systematic feature importance analysis across their forecasting models.

A second pillar of the study is its dataset. Rather than relying on synthetic traces or borrowed benchmarks, the team generated a real time-series dataset by running multiple applications in the form of containers on virtual machines. This is a meaningful design choice. Containers have become the dominant packaging unit for modern cloud applications, and their resource footprints differ from those of monolithic virtual machines. By instrumenting a live environment in which containerized applications competed for host resources, the researchers captured load dynamics that more closely resemble production conditions. The dataset has been made publicly available on GitHub, which the authors state was generated by them, giving other researchers a chance to reproduce and extend the results.

On the modeling side, the proposed deep temporal framework was benchmarked against a formidable lineup of deep learning architectures that are widely used for time-series forecasting. The comparison set included CNN-LSTM hybrids, which combine convolutional feature extraction with recurrent sequence modeling; GRU networks, a streamlined variant of the recurrent family; temporal convolutional networks, known as TCNs, which use dilated causal convolutions to capture long-range temporal dependencies; sequence-to-sequence models; autoencoders; and graph neural networks, or GNNs. Each of these architectures brings a different inductive bias to the forecasting problem, and the breadth of the comparison matters because no single architecture dominates every workload regime.

Evaluation was carried out with a standard battery of regression metrics: accuracy, mean absolute percentage error, abbreviated as MAPE, root mean square error, or RMSE, and mean square error, or MSE. These metrics probe different aspects of forecast quality. MAPE expresses error as a percentage of the true value, making it easy to interpret operationally, while RMSE and MSE penalize large misses more heavily, which is critical in cloud management where a single badly underestimated load spike can trigger throttling or outages. Across this suite, the proposed model achieved the lowest average absolute percentage error and reached roughly 91 percent predictive performance, which the authors describe as a higher level of accuracy in workload prediction compared to current cutting-edge models.

The technical significance of combining deep temporal modeling with SHAP-based explanation goes beyond leaderboard numbers. When a forecasting model reveals which features drive its predictions, operators gain actionable insight into what actually causes load on their hosts. If, for example, the explanation analysis consistently highlights certain resource counters or temporal patterns as dominant, capacity planners can prioritize monitoring those signals and design scheduling policies around them. Explainability also builds trust: cloud providers are understandably reluctant to let a black-box model automatically trigger migrations or shutdowns, but a model whose reasoning can be audited is far easier to certify for operational use. The authors’ finding that SHAP delivered consistent and transparent evaluations suggests it can serve as a reliable interpretability layer in such pipelines.

The study situates itself within a rapidly growing literature on cloud workload prediction. Prior work has explored artificial neural networks tuned with adaptive differential evolution, auto-adaptive learning for dynamic cloud environments, CNN-LSTM models for resource utilization forecasting, and uncertainty-aware predictions with transfer learning. Recent years have also seen transformer-based and attention-driven architectures migrate from natural language processing into time-series domains ranging from financial markets to wildfire spread. The new framework’s contribution is to pull two threads together that had largely run in parallel: high-performing deep temporal forecasting and post-hoc explainability, delivered in a real-time, cloud-native setting with a container-based dataset.

The energy angle deserves particular emphasis. Data centers are among the fastest-growing consumers of electricity worldwide, and the referenced literature in the paper explicitly connects information and communications technologies to sustainable development goals. Accurate short-term load forecasting is one of the levers available for greener computing: if a scheduler can anticipate which hosts will be underutilized, it can consolidate workloads and power down idle machines before they burn energy doing nothing. A forecasting framework that reaches about 91 percent predictive performance with a low percentage error could therefore translate directly into measurable energy savings, provided the predictions are fast enough to act on, which is precisely the real-time capability the framework is designed to deliver.

Limitations and open questions remain, as with any study. The dataset, while generated from real containerized applications, comes from the authors’ own experimental environment, and generalization to hyperscale production clusters with radically different workload mixes will need further validation. The authors note there was no external funding for the study and declare no conflict of interest, and all three researchers contributed equally, with Bawa handling conceptualization, methodology, programming, formal analysis, and the original draft. Still, the combination of a publicly available real-world dataset, a rigorous multi-architecture comparison, and a transparent explanation layer marks this as a notable step toward cloud platforms that can not only predict their own future load but also show their work. As operators and regulators increasingly demand accountability from automated systems, forecasting models that can explain their reasoning may prove to be the ones that actually get deployed.

Subject of Research: Explainable deep learning for real-time cloud host load forecasting

Article Title: An explainable deep temporal framework for cloud-based real-time load forecasting

Article References: An explainable deep temporal framework for cloud-based real-time load forecasting. (n.d.). https://doi.org/10.1007/s10586-026-06599-4

Image Credits: AI Generated

DOI: 10.1007/s10586-026-06599-4

Keywords: cloud computing, load forecasting, deep learning, explainable AI, SHAP, time series, containers, virtual machines, CNN-LSTM, GRU, temporal convolutional network, energy efficiency

The research, carried out by Sahani Pooja Jaiprakash, Tapas Badal and Naween Kumar at Bennett University in Greater Noida, India, addresses a problem that has stubbornly resisted classical optimization. Cloud workflows are not simply piles of independent jobs; they are directed structures in which tasks depend on the outputs of other tasks, priorities shift, and the machines available to execute them differ wildly in processing power, energy profile and price. Traditional heuristics such as list-scheduling algorithms can produce good solutions for well-behaved cases, but the authors argue they tend to be inflexible, struggling to adapt to time-varying workloads and often overlooking task priorities and inter-task dependencies altogether.

The team’s answer is a framework they call RL-LBD, short for Reinforcement Learning with Long Short-Term Memory and DBSCAN. The name hints at its three cooperating components. The first is a two-stage Long Short-Term Memory network, a type of recurrent neural architecture designed to retain information across sequences. In the RL-LBD pipeline, the LSTM serves as the scheduler’s memory: it encodes the dependency structure of incoming tasks and the profiles of the machines that might execute them, so that the system does not treat each scheduling decision as if the world had just been created. That matters because in a real workflow, scheduling task forty without knowing that it blocks tasks forty-one through forty-five is a recipe for idle processors and missed deadlines.

The second component tackles scale. A scheduler facing thousands of tasks cannot sensibly reason about each one individually, so the framework uses DBSCAN, a density-based clustering algorithm, to group tasks with similar characteristics before any ordering decision is made. DBSCAN is a natural fit here because, unlike k-means-style methods, it does not require the number of clusters to be specified in advance and can label outliers, both useful properties when workload shapes are unknown and irregular. By collapsing the decision space into clusters of related tasks, the scheduling problem becomes tractable without discarding the structure that makes workflows hard in the first place.

On top of this memory-and-clustering substrate sits the third component: a Deep Q-Network agent, the same family of algorithm that famously learned to play Atari games from raw pixels. The DQN learns a policy that dynamically prioritizes the clustered tasks, effectively deciding which group of work should move to the front of the queue under current conditions. A greedy task allocation strategy then assigns the prioritized tasks to specific virtual machines, choosing placements that minimize three objectives simultaneously: makespan, the total time to finish the entire workflow; energy consumption; and execution cost. The greedy allocation is deliberately simple, because the heavy lifting of judgment has already been done by the learned prioritization policy.

Reinforcement learning is well suited to this setting for a structural reason. Unlike supervised learning, which needs labeled examples of correct answers, an RL agent learns by acting: it schedules tasks, observes the consequences in its reward signal, and adjusts. Cloud scheduling is precisely a sequential decision problem, where an early choice constrains everything that follows, and where the environment, fluctuating workloads, changing machine availability, is non-stationary. Earlier efforts in this direction, from Q-learning-based schedulers to deep Q-network approaches for multi-objective workflow scheduling, demonstrated the promise but often treated tasks as independent entities or required retraining when workflows changed. The LSTM component in RL-LBD is the authors’ mechanism for injecting temporal and structural awareness into what would otherwise be a memoryless agent.

The reported results are where the framework earns its headline. In simulations, RL-LBD yielded improvements in energy efficiency, makespan and processing cost compared with the most recent competing approaches. Crucially, the authors did not stop at point estimates. They applied one-way analysis of variance and Fisher’s Least Significant Difference post hoc tests to confirm that the performance improvements were statistically significant across all evaluated scheduling metrics. In a field where claimed gains sometimes evaporate under statistical scrutiny, the formal significance testing lends the comparison more weight than a simple bar chart would.

The framework also showed versatility across different workflow types, which the authors interpret as evidence of its ability to enhance quality of service, improve data center efficiency and reduce operational costs. Notably, they report that the system demonstrated competitive performance in a real-world heterogeneous cloud environment, not merely in synthetic benchmarks, and that it can answer basic scheduling queries in a multi-layered, dynamic way. That last capability points toward a future in which schedulers are not static policies but interactive services that cloud operators can query about trade-offs between speed, cost and energy in real time.

The broader context makes the work timely. Data centers already consume a substantial share of global electricity, and that share is growing as AI workloads multiply. Scheduling is one of the few levers that improves all three of the cloud’s most pressing metrics, latency, energy and money, without new hardware. The literature the authors build on reflects a decade of escalating ambition: hybrid evolutionary algorithms such as HEPGA and deadline-budget-constrained ant colony optimization, multi-objective methods based on artificial bee colonies and genetic algorithms, and a steady march of reinforcement learning schedulers from early random-task schemes through parallel SARSA agents to meta-reinforcement learning approaches for multi-workflow environments. RL-LBD’s contribution is architectural: it shows that memory, clustering and value-based reinforcement learning can be composed so that each covers the others’ weaknesses.

There are, of course, the usual caveats that accompany simulation-driven scheduling research. The abstract does not report specific percentage improvements or the exact workflow benchmarks used, and the authors state that no datasets were generated or analysed during the study, meaning independent replication will depend on the community reconstructing comparable test conditions. The interplay between the LSTM’s memory horizon and the DBSCAN clustering granularity also raises questions about how the framework behaves under extreme scale or highly bursty arrivals, questions the full paper presumably explores. Still, the combination of statistical rigor, multi-objective optimization and demonstrated performance in a heterogeneous setting positions RL-LBD as a credible step toward schedulers that treat cloud workflows not as static puzzles but as living sequences to be learned. If the gains hold at production scale, the payoff could arrive quietly, in faster jobs, cooler server rooms and smaller invoices, exactly the kind of invisible victory a good scheduler is supposed to deliver.

Subject of Research: Adaptive scheduling of cloud computing workflows using LSTM-enhanced deep reinforcement learning to minimize makespan, energy consumption and execution cost

Article Title: Adaptive Cloud Workflow Scheduling via LSTM-Enhanced Deep Reinforcement Learning

Article References: Jaiprakash, S. P., Badal, T., & Kumar, N. (2026). Adaptive Cloud Workflow Scheduling via LSTM-Enhanced Deep Reinforcement Learning. Mobile Networks and Applications. https://doi.org/10.1007/s11036-026-02553-1

Image Credits: AI Generated

DOI: 10.1007/s11036-026-02553-1

Keywords: cloud computing, workflow scheduling, deep reinforcement learning, LSTM, DBSCAN clustering, Deep Q-Network, energy efficiency, makespan, virtual machines, quality of service, greedy task allocation, heterogeneous clouds