Ramin Khalili

Kilian Pfeiffer, Martin Rapp, Ramin Khalili et al.

With an increasing number of smart devices like internet of things (IoT) devices deployed in the field, offloadingtraining of neural networks (NNs) to a central server becomes more and more infeasible. Recent efforts toimprove users' privacy have led to on-device learning emerging as an alternative. However, a model trainedonly on a single device, using only local data, is unlikely to reach a high accuracy. Federated learning (FL)has been introduced as a solution, offering a privacy-preserving trade-off between communication overheadand model accuracy by sharing knowledge between devices but disclosing the devices' private data. Theapplicability and the benefit of applying baseline FL are, however, limited in many relevant use cases dueto the heterogeneity present in such environments. In this survey, we outline the heterogeneity challengesFL has to overcome to be widely applicable in real-world applications. We especially focus on the aspect ofcomputation heterogeneity among the participating devices and provide a comprehensive overview of recentworks on heterogeneity-aware FL. We discuss two groups: works that adapt the NN architecture and worksthat approach heterogeneity on a system level, covering Federated Averaging (FedAvg), distillation, and splitlearning-based approaches, as well as synchronous and asynchronous aggregation schemes.

Kilian Pfeiffer, Martin Rapp, Ramin Khalili et al.

Devices participating in federated learning (FL) typically have heterogeneous communication, computation, and memory resources. However, in synchronous FL, all devices need to finish training by the same deadline dictated by the server. Our results show that training a smaller subset of the neural network (NN) at constrained devices, i.e., dropping neurons/filters as proposed by state of the art, is inefficient, preventing these devices to make an effective contribution to the model. This causes unfairness w.r.t the achievable accuracies of constrained devices, especially in cases with a skewed distribution of class labels across devices. We present a novel FL technique, CoCoFL, which maintains the full NN structure on all devices. To adapt to the devices' heterogeneous resources, CoCoFL freezes and quantizes selected layers, reducing communication, computation, and memory requirements, whereas other layers are still trained in full precision, enabling to reach a high accuracy. Thereby, CoCoFL efficiently utilizes the available resources on devices and allows constrained devices to make a significant contribution to the FL system, increasing fairness among participants (accuracy parity) and significantly improving the final accuracy of the model.

Taylan Şahin, Ramin Khalili, Mate Boban et al.

Performance of vehicle-to-vehicle (V2V) communications depends highly on the employed scheduling approach. While centralized network schedulers offer high V2V communication reliability, their operation is conventionally restricted to areas with full cellular network coverage. In contrast, in out-of-cellular-coverage areas, comparatively inefficient distributed radio resource management is used. To exploit the benefits of the centralized approach for enhancing the reliability of V2V communications on roads lacking cellular coverage, we propose VRLS (Vehicular Reinforcement Learning Scheduler), a centralized scheduler that proactively assigns resources for out-of-coverage V2V communications \textit{before} vehicles leave the cellular network coverage. By training in simulated vehicular environments, VRLS can learn a scheduling policy that is robust and adaptable to environmental changes, thus eliminating the need for targeted (re-)training in complex real-life environments. We evaluate the performance of VRLS under varying mobility, network load, wireless channel, and resource configurations. VRLS outperforms the state-of-the-art distributed scheduling algorithm in zones without cellular network coverage by reducing the packet error rate by half in highly loaded conditions and achieving near-maximum reliability in low-load scenarios.

Jing Tan, Ramin Khalili, Holger Karl et al.

We formulate computation offloading as a decentralized decision-making problem with autonomous agents. We design an interaction mechanism that incentivizes agents to align private and system goals by balancing between competition and cooperation. The mechanism provably has Nash equilibria with optimal resource allocation in the static case. For a dynamic environment, we propose a novel multi-agent online learning algorithm that learns with partial, delayed and noisy state information, and a reward signal that reduces information need to a great extent. Empirical results confirm that through learning, agents significantly improve both system and individual performance, e.g., 40% offloading failure rate reduction, 32% communication overhead reduction, up to 38% computation resource savings in low contention, 18% utilization increase with reduced load variation in high contention, and improvement in fairness. Results also confirm the algorithm's good convergence and generalization property in significantly different environments.

Vasilios A. Siris, Adamantia Stamou, George D. Stamoulis et al.

The widespread use of AI services has raised concerns for its environmental sustainability, towards which recent studies have identified carbon emissions of AI inference as the major contributor. This paper introduces a framework for designing AI inference incentives based on the users' valuation for inference quality and latency, together with their environmental consciousness, while accounting for the tradeoff between carbon emissions and the two QoE parameters. Our approach can accommodate different tradeoffs, that depend on the size and complexity of the AI models and the allocation of resources to serve inference requests. The incentives can be offered through a practical two-tier service subscription that offers users a discount in exchange for reduced carbon emissions. The discounted service option gives the AI provider the flexibility to serve some percentage of inference requests at a lower quality and higher latency during periods of high carbon intensity.

Nikolaos Koursioumpas, Lina Magoula, Nikolaos Petropouleas et al.

Progressing towards a new era of Artificial Intelligence (AI) - enabled wireless networks, concerns regarding the environmental impact of AI have been raised both in industry and academia. Federated Learning (FL) has emerged as a key privacy preserving decentralized AI technique. Despite efforts currently being made in FL, its environmental impact is still an open problem. Targeting the minimization of the overall energy consumption of an FL process, we propose the orchestration of computational and communication resources of the involved devices to minimize the total energy required, while guaranteeing a certain performance of the model. To this end, we propose a Soft Actor Critic Deep Reinforcement Learning (DRL) solution, where a penalty function is introduced during training, penalizing the strategies that violate the constraints of the environment, and contributing towards a safe RL process. A device level synchronization method, along with a computationally cost effective FL environment are proposed, with the goal of further reducing the energy consumption and communication overhead. Evaluation results show the effectiveness and robustness of the proposed scheme compared to four state-of-the-art baseline solutions on different network environments and FL architectures, achieving a decrease of up to 94% in the total energy consumption.

Jing Tan, Ramin Khalili, Holger Karl et al.

We formulate offloading of computational tasks from a dynamic group of mobile agents (e.g., cars) as decentralized decision making among autonomous agents. We design an interaction mechanism that incentivizes such agents to align private and system goals by balancing between competition and cooperation. In the static case, the mechanism provably has Nash equilibria with optimal resource allocation. In a dynamic environment, this mechanism's requirement of complete information is impossible to achieve. For such environments, we propose a novel multi-agent online learning algorithm that learns with partial, delayed and noisy state information, thus greatly reducing information need. Our algorithm is also capable of learning from long-term and sparse reward signals with varying delay. Empirical results from the simulation of a V2X application confirm that through learning, agents with the learning algorithm significantly improve both system and individual performance, reducing up to 30% of offloading failure rate, communication overhead and load variation, increasing computation resource utilization and fairness. Results also confirm the algorithm's good convergence and generalization property in different environments.

Jing Tan, Ramin Khalili, Holger Karl

We propose a multi-agent distributed reinforcement learning algorithm that balances between potentially conflicting short-term reward and sparse, delayed long-term reward, and learns with partial information in a dynamic environment. We compare different long-term rewards to incentivize the algorithm to maximize individual payoff and overall social welfare. We test the algorithm in two simulated auction games, and demonstrate that 1) our algorithm outperforms two benchmark algorithms in a direct competition, with cost to social welfare, and 2) our algorithm's aggressive competitive behavior can be guided with the long-term reward signal to maximize both individual payoff and overall social welfare.

Marvin Illian, Ramin Khalili, Antonio A. de A. Rocha et al.

The widespread deployment of 5G networks, together with the coexistence of 4G/LTE networks, provides mobile devices a diverse set of candidate cells to connect to. However, associating mobile devices to cells to maximize overall network performance, a.k.a. cell (re)selection, remains a key challenge for mobile operators. Today, cell (re)selection parameters are typically configured manually based on operator experience and rarely adapted to dynamic network conditions. In this work, we ask: Can an agent automatically learn and adapt cell (re)selection parameters to consistently improve network performance? We present a reinforcement learning (RL)-based framework called CellPilot that adaptively tunes cell (re)selection parameters by learning spatiotemporal patterns of mobile network dynamics. Our study with real-world data demonstrates that even a lightweight RL agent can outperform conventional heuristic reconfigurations by up to 167%, while generalizing effectively across different network scenarios. These results indicate that data-driven approaches can significantly improve cell (re)selection configurations and enhance mobile network performance.

Jing Tan, Ramin Khalili, Holger Karl

The Intelligent Transportation System (ITS) environment is known to be dynamic and distributed, where participants (vehicle users, operators, etc.) have multiple, changing and possibly conflicting objectives. Although Reinforcement Learning (RL) algorithms are commonly applied to optimize ITS applications such as resource management and offloading, most RL algorithms focus on single objectives. In many situations, converting a multi-objective problem into a single-objective one is impossible, intractable or insufficient, making such RL algorithms inapplicable. We propose a multi-objective, multi-agent reinforcement learning (MARL) algorithm with high learning efficiency and low computational requirements, which automatically triggers adaptive few-shot learning in a dynamic, distributed and noisy environment with sparse and delayed reward. We test our algorithm in an ITS environment with edge cloud computing. Empirical results show that the algorithm is quick to adapt to new environments and performs better in all individual and system metrics compared to the state-of-the-art benchmark. Our algorithm also addresses various practical concerns with its modularized and asynchronous online training method. In addition to the cloud simulation, we test our algorithm on a single-board computer and show that it can make inference in 6 milliseconds.

Dimitrios Kafetzis, Ramin Khalili, Iordanis Koutsopoulos

Large Language Models (LLMs) based on autoregressive, decoder-only Transformers generate text one token at a time, where a token represents a discrete unit of text. As each newly produced token is appended to the partial output sequence, the length grows and so does the memory and compute load, due to the expanding key-value caches, which store intermediate representations of all previously generated tokens in the multi-head attention (MHA) layer. As this iterative process steadily increases memory and compute demands, layer-based partitioning in resource-constrained edge environments often results in memory overload or high inference latency. To address this and reduce inference latency, we propose a resource-aware Transformer architecture partitioning algorithm, where the partitioning decision is updated at regular intervals during token generation. The approach is myopic in that it is based on instantaneous information about device resource availability and network link bandwidths. When first executed, the algorithm places blocks on devices, and in later executions, it migrates these blocks among devices so that the sum of migration delay and inference delay remains low. Our approach partitions the decoder at the attention head level, co-locating each attention head with its key-value cache and allowing dynamic migrations whenever resources become tight. By allocating different attention heads to different devices, we exploit parallel execution of attention heads and thus achieve substantial reductions in inference delays. Our experiments show that in small-scale settings (3-5 devices), the proposed method achieves within 15 to 20 percent of an exact optimal solver's latency, while in larger-scale tests it achieves notable improvements in inference speed and memory usage compared to state-of-the-art layer-based partitioning approaches.

Nikolaos Koursioumpas, Lina Magoula, Ioannis Stavrakakis et al.

Beyond 5G and 6G networks are expected to support new and challenging use cases and applications that depend on a certain level of Quality of Service (QoS) to operate smoothly. Predicting the QoS in a timely manner is of high importance, especially for safety-critical applications as in the case of vehicular communications. Although until recent years the QoS prediction has been carried out by centralized Artificial Intelligence (AI) solutions, a number of privacy, computational, and operational concerns have emerged. Alternative solutions have surfaced (e.g. Split Learning, Federated Learning), distributing AI tasks of reduced complexity across nodes, while preserving the privacy of the data. However, new challenges rise when it comes to scalable distributed learning approaches, taking into account the heterogeneous nature of future wireless networks. The current work proposes DISTINQT, a novel multi-headed input privacy-aware distributed learning framework for QoS prediction. Our framework supports multiple heterogeneous nodes, in terms of data types and model architectures, by sharing computations across them. This enables the incorporation of diverse knowledge into a sole learning process that will enhance the robustness and generalization capabilities of the final QoS prediction model. DISTINQT also contributes to data privacy preservation by encoding any raw input data into highly complex, compressed, and irreversible latent representations before any transmission. Evaluation results showcase that DISTINQT achieves a statistically identical performance compared to its centralized version, while also proving the validity of the privacy preserving claims. DISTINQT manages to achieve a reduction in prediction error of up to 65% on average against six state-of-the-art centralized baseline solutions presented in the Tele-Operated Driving use case.

Yiannis Papageorgiou, Yannis Thomas, Alexios Filippakopoulos et al.

Federated learning (FL) operates based on model exchanges between the server and the clients, and it suffers from significant client-side computation and communication burden. Split federated learning (SFL) arises a promising solution by splitting the model into two parts, that are trained sequentially: the clients train the first part of the model (client-side model) and transmit it to the server that trains the second (server-side model). Existing SFL schemes though still exhibit long training delays and significant communication overhead, especially when clients of different computing capability participate. Thus, we propose Collaborative-Split Federated Learning~(C-SFL), a novel scheme that splits the model into three parts, namely the model parts trained at the computationally weak clients, the ones trained at the computationally strong clients, and the ones at the server. Unlike existing works, C-SFL enables parallel training and aggregation of model's parts at the clients and at the server, resulting in reduced training delays and commmunication overhead while improving the model's accuracy. Experiments verify the multiple gains of C-SFL against the existing schemes.

Kilian Pfeiffer, Mohamed Aboelenien Ahmed, Ramin Khalili et al.

In recent years, Large Language Models (LLMs) through Transformer structures have dominated many machine learning tasks, especially text processing. However, these models require massive amounts of data for training and induce high resource requirements, particularly in terms of the large number of Floating Point Operations (FLOPs) and the high amounts of memory needed. To fine-tune such a model in a parameter-efficient way, techniques like Adapter or LoRA have been developed. However, we observe that the application of LoRA, when used in federated learning (FL), while still being parameter-efficient, is memory and FLOP inefficient. Based on that observation, we develop a novel layer finetuning scheme that allows devices in cross-device FL to make use of pretrained neural networks (NNs) while adhering to given resource constraints. We show that our presented scheme outperforms the current state of the art when dealing with homogeneous or heterogeneous computation and memory constraints and is on par with LoRA regarding limited communication, thereby achieving significantly higher accuracies in FL training.

Theodora Panagea, Nikolaos Koursioumpas, Lina Magoula et al.

Progressing toward a new generation of mobile networks, a clear focus on integrating distributed intelligence across the system is observed to drive performance, autonomy, and real-time adaptability. Federated learning (FL) stands out as a key emerging technique, enabling on-device model training while preserving data locality. However, its operation introduces substantial energy and resource demands. Energy needs are mostly met by grid power sources, while FL resource orchestration strategies remain limited. This work introduces GreenFLag, an agentic resource orchestration framework designed to minimize the energy consumption from the grid power to complete FL workflows, guarantee FL model performance, and reduce grid power reliance by incorporating renewable sources into the system. GreenFLag leverages a Soft-Actor Critic reinforcement learning approach to jointly optimize computational and communication resources, while accounting for communication contention and the dynamic availability of renewable energy. Evaluations using a real-world open dataset from Copernicus, demonstrate that GreenFLag significantly reduces grid energy consumption by 94.8% on average, compared to three state-of-the-art baselines, while primarily relying on green power.

Mohamed Aboelenien Ahmed, Kilian Pfeiffer, Heba Khdr et al.

Training on edge devices poses several challenges as these devices are generally resource-constrained, especially in terms of power. State-of-the-art techniques at the device level reduce the GPU frequency to enforce power constraints, leading to a significant increase in training time. To accelerate training, we propose to jointly adjust the system and application parameters (in our case, the GPU frequency and the batch size of the training task) while adhering to the power constraints on devices. We introduce a novel cross-layer methodology that combines predictions of batch size efficiency and device profiling to achieve the desired optimization. Our evaluation on real hardware shows that our method outperforms the current baselines that depend on state of the art techniques, reducing the training time by $2.4\times$ with results very close to optimal. Our measurements also indicate a substantial reduction in the overall energy used for the training process. These gains are achieved without reduction in the performance of the trained model.

Mohamed Aboelenien Ahmed, Kilian Pfeiffer, Ramin Khalili et al.

Federated fine-tuning offers a promising approach for tuning Large Language Models (LLMs) on edge devices while preserving data privacy. However, fine-tuning these models on edge devices remains challenging due to high memory, communication, and computational demands. Zero-order optimization with task alignment provides a potential solution, enabling fine-tuning with inference-level memory requirements but requires a longer convergence time. In this paper, we propose Federated Split-Perturbation Zero-order Optimization (FedSPZO) that divides the network into two blocks, applying a different number of perturbations per block in a computationally effective way, achieving faster convergence. Our evaluation shows a $2.5 - 7\times $ reduction in computation overhead compared to zero-order state of the art techniques in federated learning.

Kilian Pfeiffer, Ramin Khalili, Jörg Henkel

Federated learning (FL) is usually performed on resource-constrained edge devices, e.g., with limited memory for the computation. If the required memory to train a model exceeds this limit, the device will be excluded from the training. This can lead to a lower accuracy as valuable data and computation resources are excluded from training, also causing bias and unfairness. The FL training process should be adjusted to such constraints. The state-of-the-art techniques propose training subsets of the FL model at constrained devices, reducing their resource requirements for training. But these techniques largely limit the co-adaptation among parameters of the model and are highly inefficient, as we show: it is actually better to train a smaller (less accurate) model by the system where all the devices can train the model end-to-end, than applying such techniques. We propose a new method that enables successive freezing and training of the parameters of the FL model at devices, reducing the training's resource requirements at the devices, while still allowing enough co-adaptation between parameters. We show through extensive experimental evaluation that our technique greatly improves the accuracy of the trained model (by 52.4 p.p.) compared with the state of the art, efficiently aggregating the computation capacity available on distributed devices.

Philipp Wiesner, Ramin Khalili, Dennis Grinwald et al.

Federated Learning (FL) is an emerging machine learning technique that enables distributed model training across data silos or edge devices without data sharing. Yet, FL inevitably introduces inefficiencies compared to centralized model training, which will further increase the already high energy usage and associated carbon emissions of machine learning in the future. One idea to reduce FL's carbon footprint is to schedule training jobs based on the availability of renewable excess energy that can occur at certain times and places in the grid. However, in the presence of such volatile and unreliable resources, existing FL schedulers cannot always ensure fast, efficient, and fair training. We propose FedZero, an FL system that operates exclusively on renewable excess energy and spare capacity of compute infrastructure to effectively reduce a training's operational carbon emissions to zero. Using energy and load forecasts, FedZero leverages the spatio-temporal availability of excess resources by selecting clients for fast convergence and fair participation. Our evaluation, based on real solar and load traces, shows that FedZero converges significantly faster than existing approaches under the mentioned constraints while consuming less energy. Furthermore, it is robust to forecasting errors and scalable to tens of thousands of clients.

Martin Rapp, Ramin Khalili, Kilian Pfeiffer et al.

We study the problem of distributed training of neural networks (NNs) on devices with heterogeneous, limited, and time-varying availability of computational resources. We present an adaptive, resource-aware, on-device learning mechanism, DISTREAL, which is able to fully and efficiently utilize the available resources on devices in a distributed manner, increasing the convergence speed. This is achieved with a dropout mechanism that dynamically adjusts the computational complexity of training an NN by randomly dropping filters of convolutional layers of the model. Our main contribution is the introduction of a design space exploration (DSE) technique, which finds Pareto-optimal per-layer dropout vectors with respect to resource requirements and convergence speed of the training. Applying this technique, each device is able to dynamically select the dropout vector that fits its available resource without requiring any assistance from the server. We implement our solution in a federated learning (FL) system, where the availability of computational resources varies both between devices and over time, and show through extensive evaluation that we are able to significantly increase the convergence speed over the state of the art without compromising on the final accuracy.

Martin Rapp, Ramin Khalili, Jörg Henkel

We consider a distributed system, consisting of a heterogeneous set of devices, ranging from low-end to high-end. These devices have different profiles, e.g., different energy budgets, or different hardware specifications, determining their capabilities on performing certain learning tasks. We propose the first approach that enables distributed learning in such a heterogeneous system. Applying our approach, each device employs a neural network (NN) with a topology that fits its capabilities; however, part of these NNs share the same topology, so that their parameters can be jointly learned. This differs from current approaches, such as federated learning, which require all devices to employ the same NN, enforcing a trade-off between achievable accuracy and computational overhead of training. We evaluate heterogeneous distributed learning for reinforcement learning (RL) and observe that it greatly improves the achievable reward on more powerful devices, compared to current approaches, while still maintaining a high reward on the weaker devices. We also explore supervised learning, observing similar gains.

Taylan Şahin, Ramin Khalili, Mate Boban et al.

Vehicle-to-vehicle (V2V) communications have distinct challenges that need to be taken into account when scheduling the radio resources. Although centralized schedulers (e.g., located on base stations) could be utilized to deliver high scheduling performance, they cannot be employed in case of coverage gaps. To address the issue of reliable scheduling of V2V transmissions out of coverage, we propose Vehicular Reinforcement Learning Scheduler (VRLS), a centralized scheduler that predictively assigns the resources for V2V communication while the vehicle is still in cellular network coverage. VRLS is a unified reinforcement learning (RL) solution, wherein the learning agent, the state representation, and the reward provided to the agent are applicable to different vehicular environments of interest (in terms of vehicular density, resource configuration, and wireless channel conditions). Such a unified solution eliminates the necessity of redesigning the RL components for a different environment, and facilitates transfer learning from one to another similar environment. We evaluate the performance of VRLS and show its ability to avoid collisions and half-duplex errors, and to reuse the resources better than the state of the art scheduling algorithms. We also show that pre-trained VRLS agent can adapt to different V2V environments with limited retraining, thus enabling real-world deployment in different scenarios.

Taylan Şahin, Ramin Khalili, Mate Boban et al.

Radio resources in vehicle-to-vehicle (V2V) communication can be scheduled either by a centralized scheduler residing in the network (e.g., a base station in case of cellular systems) or a distributed scheduler, where the resources are autonomously selected by the vehicles. The former approach yields a considerably higher resource utilization in case the network coverage is uninterrupted. However, in case of intermittent or out-of-coverage, due to not having input from centralized scheduler, vehicles need to revert to distributed scheduling. Motivated by recent advances in reinforcement learning (RL), we investigate whether a centralized learning scheduler can be taught to efficiently pre-assign the resources to vehicles for out-of-coverage V2V communication. Specifically, we use the actor-critic RL algorithm to train the centralized scheduler to provide non-interfering resources to vehicles before they enter the out-of-coverage area. Our initial results show that a RL-based scheduler can achieve performance as good as or better than the state-of-art distributed scheduler, often outperforming it. Furthermore, the learning process completes within a reasonable time (ranging from a few hundred to a few thousand epochs), thus making the RL-based scheduler a promising solution for V2V communications with intermittent network coverage.

Yung-Chih Chen, Don Towsley, Ramin Khalili

Online video streaming through mobile devices has become extremely popular nowadays. YouTube, for example, reported that the percentage of its traffic streaming to mobile devices has soared from 6% to more than 40% over the past two years. Moreover, people are constantly seeking to stream high quality video for better experience while often suffering from limited bandwidth. Thanks to the rapid deployment of content delivery networks (CDNs), popular videos are now replicated at different sites, and users can stream videos from close-by locations with low latencies. As mobile devices nowadays are equipped with multiple wireless interfaces (e.g., WiFi and 3G/4G), aggregating bandwidth for high definition video streaming has become possible. We propose a client-based video streaming solution, MSPlayer, that takes advantage of multiple video sources as well as multiple network paths through different interfaces. MSPlayer reduces start-up latency and provides high quality video streaming and robust data transport in mobile scenarios. We experimentally demonstrate our solution on a testbed and through the YouTube video service.