OMNIS+: Semantic RAN Slicing and Quantization for Dynamic Split Neural Networks
Authors: Langtian Qin, Avishkar Dnyaneshwar Kawadgave, Ian Andrew Harshbarger, Leïla Nasraoui, Carla Fabiana Chiasserini, Marco Levorato
Organizations: Donald Bren School of Information and Computer Sciences, University of California, Irvine, USA · National School of Computer Science (ENSI), University of Manouba, Tunisia · Politecnico di Torino, Italy · CNIT, Parma, Italy · Chalmers University of Technology, Sweden
Edge computing enables resource-constrained devices to execute machine learning (ML) applications via task offloading, with radio access network (RAN) slicing playing a crucial role in provisioning joint communication and computing resources. However, existing RAN slicing frameworks rely on static computing models, limiting their ability to exploit dynamic semantic data representations enabled by modern neural architectures. To address this limitation, we propose OMNIS+, a semantic RAN slicing framework for multi-cell edge computing built upon a new generation of dynamic split neural models. OMNIS+ incorporates dynamic neural compression paired with adaptive quantization for task offloading, providing a versatile decision space across communication payload and compute dimensions. We leveraged a novel ``box'' quantization scheme that markedly enhances resilience to transmission bit errors across diverse compression rates. Considering partial system observability and distinct node objectives, we formulate two interdependent optimization problems: (i)~mobile devices (MDs) maximize inference accuracy under quality-of-service (QoS) and queue-stability constraints by jointly adapting dynamic split deep neural network (DNN) branches and cell association; and (ii)~each edge server (ES) allocates bandwidth, graphics processing unit (GPU) frequency, and modulation and coding scheme (MCS) to maximize the max--min inference accuracy among associated MDs under cell capacity limits. We solve these problems using a multi-agent distributed optimization framework, where MDs act as causal contextual multi-armed bandit (MAB) agents under backlog-aware Lyapunov control, while each ES executes resource allocation via block coordinate descent (BCD). Compared with state-of-the-art baselines, OMNIS+ reduces the QoS violation probability by up to 1.75 while achieving a better accuracy--violation trade-off.
Figures & tables
Fig. 1: Comparison of existing RAN slicing frameworks for edge computing and OMNIS+: (a) Offloading with neural network partitioning [ 3 ] , (b) Offloading with static data compression [ 8 , 9 ] , and (c) OMNIS+.
Fig. 2: Accuracy vs. payload size under STD quantization at SNR equal to 3 dB (left) and 10 dB (right).
Fig. 3: System model of OMNIS+. Randomly arriving tasks are queued at each MD. When a task is admitted, the MD selects a dynamic split-DNN branch and a Top- L serving cell and keeps that association until the task fully completes. Concurrent edge-stage tasks share the cell GPU-frequency pool at the co-located BS/ES. Per-cell RAN control allocates bandwidth, GPU frequency, and MCS.
Fig. 4: Accuracy vs. payload size for different quantization methods at various SNRs: 3 dB top-left, 5 dB top-right, 7 dB bottom-left, 10 dB bottom-right.
Fig. 5: Two-timescale OMNIS+ control flow.
Fig. 6: Convergence and running time performance of different schemes. The top panel shows the Lyapunov reward per slot averaged over MDs over 500 slots. The bottom panel shows control plane running time.
Fig. 7: Performance of different schemes under varying SNR. The top row shows reward, average delay, and energy from left to right. The bottom row shows accuracy, violation probability, and backlog from left to right.
Fig. 8: Performance of different schemes under varying numbers of MDs. The top row shows reward, average delay, and energy from left to right. The bottom row shows accuracy, violation probability, and backlog from left to right.
Fig. 9: Performance of different schemes under varying task arrival rates. The top row shows reward, average delay, and energy from left to right. The bottom row shows accuracy, violation probability, and backlog from left to right.
Fig. 10: Action pick probabilities of different schemes under varying parameters. The top row corresponds to SNR =2,4,6 dB from left to right. The bottom row corresponds to M=10,15,20 from left to right.
Fig. 11: Accuracy and violation trade-off of OMNIS+ and GDO.
Radio Access Network (RAN) slicing is a shared resource allocation problem: each slice seeks to satisfy its Service Level Agreement (SLA) with minimal radio-resource consumption, while all slices compete for a limited resource budget. This creates both an intra-slice trade-off between SLA assurance and resource efficiency and an inter-slice conflict over shared resources, calling for specialized yet coordinated control. In this work, we present CoSlice, a coordinated multi-agent reinforcement learning framework for inter-slice physical resource block allocation in Open Radio Access Network (O-RAN). CoSlice uses preference-conditioned slice agents for fast resource allocation and an LLM-based coordinator to adapt their objectives using global network observations at a slower timescale. We formulate the problem as a multi-objective partially observable stochastic game (mo-POSG), establish the existence of a Preference Correlated Equilibrium, and derive a sublinear regret bound for the learning procedure. We implement the algorithm by using a two-stage training procedure that consists of multi-objective pretraining and joint optimization. We build a network slicing evaluation environment using ORANSlice and its radio-frequency simulator (RFsim), supporting realistic protocol-stack execution and application traffic across multiple slices. Experimental results show that CoSlice reduces both average and tail SLA violations compared with other centralized and multi-agent baselines under diverse slice workloads.
Pengxin Wang, Nakjung Choi, Jingdi Chen
Department of Electrical and Computer Engineering, University of Arizona · Nokia & Bell Labs
Split computing constitutes a widely used distributed inference approach, where a lightweight head model is onloaded onto the device and a heavier tail model resides on an edge server, leveraging the growing computational capabilities of modern System-on-Chips while alleviating server load. As intelligent indoor environments such as smart offices grow increasingly populated with diverse IoT devices, a single edge server must simultaneously assist multiple devices, each competing for the same shared inference resources. Without a principled mechanism to manage this shared load, the server is quickly overwhelmed, causing latency SLO violations and rendering server-assisted inference ineffective. In this work, we present MANE, a distributed inference framework that equips the server with a multi-path tail architecture, enabling a dynamic accuracy--throughput trade-off at runtime. By introducing a novel multi-path model architecture, a three-stage training scheme featuring a Joint Head Network Distillation loss and a hysteresis-based scheduler with an equitable device-fallback policy, MANE maintains over 80% SLO satisfaction rate where state-of-the-art onloading methods fail completely, while preserving accuracy 6pp higher than on-device alternatives, across up to 40 concurrent devices.
Sokratis Nikolaidis, Stylianos I. Venieris, Leonidas Malachias +1
National Technical University of Athens, Athens, Greece · Samsung AI Center, Cambridge, UK
Deploying state-of-the-art deep neural networks (DNNs) at the wireless edge is severely bottlenecked by the strict energy and resource constraints of mobile devices. While federated split learning (FSL) mitigates on-device computation by offloading workloads to an edge server, this may introduce systemic overheads, while the continuous exchange of cut-layer data, and submodels still incurs significant energy consumption (EC). To address this, we propose a green quantized FSL (GQ-FSL) framework that incorporates stochastic quantization for both local collaborative training and wireless transmissions. Notably, GQ-FSL supports asymmetric precision levels for the client- and server-side submodels, effectively decoupling device energy constraints from global convergence degradation. To quantify these tradeoffs, we develop parameterized energy models for the split architecture and derive a theoretical convergence bound under statistically heterogeneous data. Building on that, we formulate a joint optimization problem to configure the DNN split point and precision levels, minimizing the total system EC while satisfying a strict target accuracy constraint. Ultimately, we demonstrate that GQ-FSL enables large-scale DNN deployment on resource-constrained devices, achieving superior energy efficiency compared to quantized federated learning and full-precision FSL.
Idan Roth, Lutz Lampe
The University of British Columbia, Vancouver, BC, Canada