WorksComputer Science
SliceWise: Traffic-Aware Multi-Agent RL with Temporal Encoding for Joint Slice Admission and Resource Orchestration in O-RAN
Abstract
Ultra-reliable low-latency communication (URLLC) services in mission-critical applications such as industrial automation, remote healthcare, and autonomous systems demand stringent quality-of-service (QoS) guarantees under dynamic, resource-constrained wireless conditions. Open radio access networks (O-RAN) offer a software-defined and programmable foundation for embedding AI-driven control. This letter proposes SliceWise, a traffic-aware multi-agent deep reinforcement learning (MARL) framework for joint admission control, user-slice association, and resource orchestration in O-RAN. The control problem is modeled as a cooperative multi-agent Markov decision process, with long short-term memory (LSTM) networks capturing temporal traffic dynamics and dueling double deep Q-networks (D3QN) ensuring stable learning. Simulation results show that SliceWise substantially reduces delay violations and packet drops while improving slice-level reliability and resource efficiency, offering a scalable orchestration solution for AI-native O-RAN deployments.
Introduction
The proliferation of latency-sensitive and mission-critical applications in 5G and beyond, including industrial automation, remote healthcare, and connected autonomous systems, imposes stringent quality-of-service (QoS) requirements on radio access networks (RAN) [1]. Ultra-reliable low-latency communication (URLLC) demands millisecond-level end-to-end latency and extremely high reliability under dynamic and bursty traffic conditions [2]. Ensuring these guarantees in the presence of limited spectrum resources, heterogeneous service demands, and fluctuating network loads is challenging, particularly in network slicing environments where maintaining service-level agreements (SLAs) across slices during congestion is complex.
Open RAN (O-RAN) provides a disaggregated, software-defined architecture with programmable interfaces that enable fine-grained control of heterogeneous services [3]. The near-real-time RAN Intelligent Controller (near-RT RIC) facilitates data-driven decision-making for admission control, user-slice association, and resource orchestration [4]. While these capabilities position O-RAN as an ideal platform for embedding AI-driven control, existing resource and admission control mechanisms remain inadequate for URLLC. Static bandwidth partitioning, heuristic scheduling, and centralized optimization [5] often lack adaptability to bursty traffic, and recent machine learning (ML) [6] and reinforcement learning (RL) [7, 8] approaches typically treat admission control, slice association, and resource orchestration in isolation, neglecting their interdependence and the distributed nature of O-RAN.
This letter proposes SLICEWISE, a traffic-aware multi-agent deep reinforcement learning (MARL) framework for joint admission control, slice association, and resource orchestration in O-RAN. The problem is modeled as a cooperative multi-agent Markov decision process (MMDP), with agents specialized for each control function. Long short-term memory (LSTM) networks capture temporal traffic dynamics, and dueling double deep Q-networks (D3QN) provide stable and efficient policy learning in high-dimensional, non-stationary environments. The main contributions are as follows:
We propose a unified MARL framework that jointly optimizes admission control, slice association, and resource allocation for O-RAN;
We develop LSTM-based agent designs that exploit temporal features and balance URLLC reliability with resource efficiency;
Extensive simulations show that SLICEWISE improves delay, reliability, and efficiency compared to random, empirical, and partially coordinated RL baselines.
System Model
We consider an O-RAN architecture supporting network slicing for differentiated service provisioning as illustrated in Figure 1. Let denote the set of user equipments (UEs) served by an O-RAN radio unit (O-RU) under a centralized unit (O-CU), which coordinates with the distributed unit (O-DU) to allocate physical resource blocks (PRBs).

Figure 1. Integrated admission control and resource orchestration framework in O-RAN architecture.
Inter-Slice Resource Partitioning and Admission Control
The system operates over a finite time horizon , with each time slot of fixed duration . To support dynamic and SLA-driven traffic management, the RAN is logically partitioned into a set of active slices , each designed to serve a specific class of traffic. Resource allocation decisions are made at discrete reconfiguration windows, indexed by , where denotes the duration (in slots) of each window. The reconfiguration window starts at time and spans the time slots , with to ensure non-overlapping intervals. At the beginning of each window, the near-RT RIC performs user admission control and inter-slice PRB partitioning. Each slice is assigned a resource budget , constrained by the total PRBs, i.e., , . Let denote the set of UEs requesting admission at the start of window , and denote the currently admitted users, with and . The binary admission decision for UE is represented as , where 1 denotes admission. The complete admission vector is defined as . The set of newly admitted UEs is denoted as , and its cardinality is defined as . The set of active UEs is then updated as , and remains unchanged throughout the set of time slots , i.e., . User-slice association is determined by a binary matrix , where assigns UE to exactly one slice, i.e., , . Accordingly, the UEs associated with slice at window are defined as with .
Intra-Slice Resource Allocation and Service Reliability
Each UE is located at a time-varying distance from the O-RU at time slot . The average path loss for UE at time is modeled as a weighted combination of line-of-sight (LoS) and non-LoS components [9] as , where is the LoS probability. The free-space path loss (FSPL) and non-LoS path loss (PL) components are computed as follows:
where is the carrier frequency, and , represent the excess attenuation for LoS and non-LoS links, respectively. Let denote the end-to-end transmission delay for UE at time , which is computed as follows:
where is the packet size, is the number of packets, and is the achievable data given by:
with signal-to-interference-plus-noise ratio (SINR) for UE at time defined as , where denotes the transmit power of the O-RU, is the number of PRBs allocated to UE , is the bandwidth per PRB, is the fading channel gain, is the noise power spectral density, and is the interference power. The intra-slice resource constraint ensures that for each slice and time . To meet SLA requirements, each UE must satisfy the reliability condition defined as , , where corresponds to latency bounds for enhanced multimedia broadband (eMBB) () and URLLC () services, and is the fulfillment threshold.
Problem Formulation
Slice reliability measures the fraction of admitted UEs whose latency remains within SLA thresholds. It is defined at window for slice as:
where for all and is a smooth, monotonically increasing function quantifying the satisfaction margin of the latency requirement defined as
where controls the slope and adjusts the offset to reflect sensitivity to slice-specific reliability. To ensure temporal reliability, we define a long-term utility that combines instantaneous and discounted future performance, jointly maximizing user admission and slice reliability at time :
where is a discount factor and represents the product of admission and slice reliabilities across future windows, expressed as , . A penalty term captures current and future PRB utilization across admitted users to enforce resource efficiency:
The overall optimization problem is thus formulated as:
s.t.
where is a scalar balancing the trade-off between performance and efficiency, and is the minimum reliability threshold for slice . The constraint in (11) enforces the expected long-term reliability of each slice, defined as
with arguments of implicitly shifting over future windows as in (6).
Methodology & Approach
To coordinate admission, association, and orchestration under dynamic conditions, we model the system as an MMDP and develop SLICEWISE, a DRL framework combining LSTM encoders with a D3QN for temporally aware control.
MMDP Framework and Agent Design
The proposed system consists of a set of agents , where “ac” denotes the admission control agent, “sa” the slice association agent, and “res” the resource orchestrator for slice . The MMDP is defined by the tuple , where represents the global environment state, and are the observation and action spaces of agent , is the transition model, is the reward function, and is the discount factor. Each agent follows a policy , selecting actions based on local observations . Agents receive feedback from the environment through rewards , guiding policy updates over time.
Observations and Action Spaces
Admission Control Agent
At the beginning of each time window , the admission control agent observes a compact summary of the network’s current load and quality conditions. This includes available bandwidth, recent delay-violation trends, aggregate queue statistics, and average channel quality across both admitted and requesting UEs. Formally, its observation vector is given by:
where averaged system delay violation rate within a time window . Similarly, the average data processing requests from admitted users and new applicants are respectively and . To maintain system feasibility, the maximum number of admissible UEs per time slot is constrained to . Based on this observation, the agent selects an action , where each action represents admitting a subset of candidate UEs according to the learned policy that balances SINR conditions, queue occupancy, and latency satisfaction.
Resource Orchestration Agent
The orchestration agent operates at a finer timescale to adjust PRB allocations for each slice during every slot . Its observation combines delay bounds, SINR, traffic load, and current slice allocation:
From this observation, the agent selects , corresponding to PRB adjustment factors that scale the slice-level allocation . The resulting per-UE bandwidth update is:
where estimates per-user bandwidth demand. To ensure resource allocation remains within capacity limits, we employ the minimum adjustment method.
Slice Association Agent
The association agent observes recent traffic sequences of all active users to infer their volatility patterns: , where,
The parameter defines the number of past time windows considered, thereby controlling the temporal horizon over which traffic variability is captured. The resulting association matrix is defined as in Section II.
Reward Functions
The admission control agent’s reward function is formulated to optimize user admissions while ensuring stringent reliability constraints, which mathematically is defined as follows:
where function . Average transmission delay over a time slot is calculated as and is a scaling factor. For the resource agent in charge of the orchestration of slice , the reward at time slot reflects system performance in terms of delay violation and resource efficiency defined as:
where and .
Learning and Training Strategy
Training proceeds over multiple episodes of time slots. At episode start, UEs are randomly assigned eMBB or URLLC service types with corresponding delay constraints. Traffic dynamics, including packet count , packet size , and distance , evolve via Markov chains with transition matrices , where indexes variable type and the traffic profile. Transitions are restricted to adjacent states to ensure smooth variability.
Each D3QN-based agent maintains separate target and online networks, with experience replay buffers of size storing tuples . Parameters include learning rate
, discount factor , soft update factor , and batch size . Admission and association agents update at each reconfiguration window, while resource agents update every slot. An -greedy policy with decaying guides exploration.
User–slice association is implemented as an LSTM classifier trained offline on traffic sequences with volatility1 labels and kept fixed during RL. This preserves a stable mapping from temporal traffic patterns to slice labels, reduces non-stationarity in the control loop, and meets near-RT inference constraints. At runtime, the LSTM outputs slice mappings, followed by inter-slice resource allocation, i.e.,
A warm-up period of slots precedes policy execution.
Simulation Results and Analysis
The proposed SLICEWISE framework is implemented and evaluated on a customized Python-based version of the open-source AI-RAN Simulator [10]. The setup emulates a multi-cell O-RAN with a serving O-RU and two interfering O-RUs positioned 2.5 and 3 km apart, all operating on the same 2.5 GHz carrier. Inter-cell interference is incorporated in each UE’s SINR. The total bandwidth under carrier aggregation MHz is discretized into PRBs with subcarrier spacing kHz. Key parameters and learning hyperparameters appear in Table I. Each simulation episode consists of time slots of duration s. Training is conducted for episodes, where the orchestration agent acts at every slot while admission and slice-association agents update once every slots. Results are averaged over
| Parameters | Value | Parameters | Value |
| (m) | , | , | |
| (Mb) | |||
| (dBm/Hz) | , (dB) | , | |
| (s) | (s) | ||
Table I. Simulation Parameters.
The network accommodates active slices and UEs distributed among them. Each UE’s traffic process is governed by a first-order Markov chain, capturing temporal correlation in both packet arrival count and packet size . For each slice, the instantaneous values of these variables are drawn uniformly from pre-defined intervals, with transitions constrained by a probability of 0.5, allowing either self-transition or progression to the next volatility state. To systematically examine framework adaptability, slices are configured with escalating volatility levels.
Slice 1 (benign) operates with narrow, stable ranges: for eMBB, MB and ; for URLLC, Kb and . This variability is incrementally expanded across slices. For example, Slice 4 (highly volatile) uses MB and for eMBB, and Kb with for URLLC. While traditional slice assignment relies on predefined QoS classes, we infer association dynamically from observed traffic volatility, enabling adaptive alignment between user behavior and slice characteristics.
As shown in Figure 2(a)–(Figure 2b), cumulative rewards for admission and orchestration agents increase monotonically, indicating stable convergence. Delay-violation and packet-drop rates (Figure 2(c)–(Figure 2d)) decline rapidly, with the traffic-aware configuration (TA–MAROAC) converging faster and outperforming MARO and MAROAC, confirming the advantage of temporal modeling and joint optimization. The trained policy is tested on unseen multi-cell episodes with variable bandwidth and heterogeneous traffic. Compared with Rand, Empirical, MARO, and MAROAC, SLICEWISE achieves the highest reliability and lowest delay violations (Figure 3(a)–(Figure 3b)). Slice-level results (Figure 3(c)–(Figure 3d)) show sustained URLLC latency compliance even under interference.

Figure 2. Evolution of training metrics – (a) the reward of admission control agent, (b) the reward of resource allocation agent, (c) the delay violation rate, and (d) packet drop rate.

Figure 3. Comparative evaluation results varying in MHz – (a) system reliability, (b) system delay violation (c) slice-wise reliability, and (d) slice-wise delay violation.
Fig. 4(a) demonstrates that SLICEWISE maintains efficient PRB utilization without over-provisioning. Although the Empirical scheme slightly reduces packet-drop rates (Figure 4(b)), its ECCDF delay profiles in Figure 4(c)–(Figure 4d) exhibit long tails, indicating SLA violations. In contrast, SLICEWISE yields compact delay distributions for both eMBB and URLLC slices, reflecting stable queue dynamics and superior control over latency dispersion. As shown in Figure 5(a), SLICEWISE achieves the best joint optimization of user admission, SLA fulfillment, and resource utilization. The sensitivity analysis with respect to the latency fulfillment threshold (Figure 5(b)) shows that all methods degrade under stricter delay requirements, but SLICEWISE experiences slowest decline showing robustness and adaptability across varying conditions.

Figure 4. Comparative evaluation results varying in MHz – (a) system resource utilization, (b) system packet drop rate (c) ECCDF of delay of URLLC users, and (d) ECCDF of delay of eMBB users.

Figure 5. Resource efficiency evaluation varying in MHz-(a) resource efficiency under various orchestration schemes, (b) resource efficiency of the proposed scheme varying fulfillment threshold .
Conclusion
This paper introduced SLICEWISE a traffic-aware multi-agent reinforcement learning framework that seamlessly integrates resource orchestration and admission control within O-RAN environments. To address the strict slice-specific requirements of URLLC and eMBB traffic, our approach combines a multi-agent deep reinforcement learning model with an LSTM network to dynamically manage user admissions, slice associations, and resource allocations. This dual-model strategy enables real-time optimization of latency, reliability, and resource efficiency while effectively balancing multiple objectives amid evolving network conditions. Notably, the D3QN-based agents in our framework exhibit robust performance, achieving significant reductions in delay violation and packet drop rates compared to baseline methods.
References
- [1]P. Rost, C. Mannweiler, D. S. Michalopoulos et al., “Network slicing to enable scalability and flexibility in 5G mobile networks,” IEEE Commun. Mag., vol. 55, no. 5, pp. 72–79, 2017.arxiv.org/abs/1704.02129
- [2]R. Han et al., “Dynamic network slice for bursty edge traffic,” IEEE/ACM Transactions on Networking, vol. 32, no. 4, pp. 3142–3157, 2024.
- [3]M. Polese, L. Bonati, S. D’Oro et al., “Understanding O-RAN: Architecture, interfaces, algorithms, security, and research challenges,” IEEE Commun. Surveys Tuts., vol. 25, no. 2, pp. 1376–1411, 2023.arxiv.org/abs/2202.01032
- [4]H.-T. Thieu, V.-Q. Pham, A. Kak, and N. Choi, “Demystifying the near-real-time RIC: Architecture, operations, and benchmarking insights,” in IEEE Conf. on Comput. Commun. Wkshps. (INFOCOM Wkshps 2023), Hoboken, NJ USA, May 2023.DOI
- [5]T. Wang and S. Wang, “Online convex optimization for efficient and robust inter-slice radio resource management,” IEEE Transactions on Communications, vol. 69, no. 9, pp. 6050–6062, 2021.DOI
- [6]N. Apostolakis, M. Gramaglia, L. E. Chatzieleftheriou, T. Subramanya, A. Banchs, and H. Sanneck, “ATHENA: machine learning and reasoning for radio resources scheduling in vRAN systems,” IEEE Journal on Selected Areas in Communications, vol. 42, no. 2, pp. 263–279, 2024.DOI
- [7]M. Yin, Y. Deng, A. Kak, N. Choi, and T. Han, “PSASlicing: Perpetual SLA-aware reinforcement learning for O-RAN slice management,” in IEEE Global Communications Conference (GLOBECOM 2024), Cape Town, South Africa, Dec. 2024, pp. 4534–4539.DOI
- [8]M. Zangooei, M. Golkarifard, M. Rouili, N. Saha, and R. Boutaba, “Flexible RAN slicing in open RAN with constrained multi-agent reinforcement learning,” IEEE Journal on Selected Areas in Communications, vol. 42, no. 2, pp. 280–294, 2024.DOI
- [9]K. Hanada, J. Zhang et al., “5G 3GPP-like channel models for outdoor urban microcellular and macrocellular environments,” in IEEE 83rd Vehicular Technology Conference (VTC Spring 2016), 2016, pp. 1–7.
- [10]A. Ntung, M. S. Rahman, T. Saha et al., “AI-RAN Simulator: Open-source simulator for AI-native O-RAN resource management,” https://github.com/ntutangyun/ai-ran-sim, 2023, accessed: Oct. 7, 2025.