Abstract Offline model-based reinforcement learning (MBRL) serves as a competitive framework that can learn well-performing policies solely from pre-collected data with the help of learned dynamics models. To fully unleash the power of offline MBRL, model selection plays a pivotal role in determining the dynamics model utilized for downstream policy learning. However, offline MBRL conventionally relies on validation or off-policy evaluation, which are rather inaccurate due to the inherent distribution shift in offline RL. To tackle this, we propose BOMS, an active model selection framework that enhances model selection in offline MBRL with only a small online interaction budget, through the lens of Bayesian optimization (BO). Specifically, we recast model selection as BO and enable probabilistic inference in BOMS by proposing a novel model-induced kernel, which is theoretically grounded and computationally efficient. Through extensive experiments, we show that BOMS improves over the baseline methods with a small amount of online interaction comparable to only $$1\%$$ - $$2.5\%$$ of offline training data on various RL tasks.

Kuang-Da Wang, Shuoyang Ding, Chao-Han Huck Yang, Ping-Chun Hsieh, Wen-Chih Peng, Vitaly Lavrukhin, Boris Ginsburg. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Bayesian optimization (BO) offers an efficient pipeline for optimizing black-box functions with the help of a Gaussian process prior and an acquisition function (AF). Recently, in the context of single-objective BO, learning-based AFs witnessed promising empirical results given its favorable non-myopic nature. Despite this, the direct extension of these approaches to multi-objective Bayesian optimization (MOBO) suffer from the hypervolume identifiability issue, which results from the non-Markovian nature of MOBO problems. To tackle this, inspired by the non-Markovian RL literature and the success of Transformers in language modeling, we present a generalized deep Q-learning framework and propose BOFormer, which substantiates this framework for MOBO via sequence modeling. Through extensive evaluation, we demonstrate that BOFormer constantly achieves better performance than the benchmark rule-based and learning-based algorithms in various synthetic MOBO and real-world multi-objective hyperparameter optimization problems.

Action-constrained reinforcement learning (ACRL) is a generic framework for learning control policies with zero action constraint violation, which is required by various safety-critical and resource-constrained applications. The existing ACRL methods can typically achieve favorable constraint satisfaction but at the cost of either high computational burden incurred by the quadratic programs (QP) or increased architectural complexity due to the use of sophisticated generative models. In this paper, we propose a generic and computationally efficient framework that can adapt a standard unconstrained RL method to ACRL through two modifications: (i) To enforce the action constraints, we leverage the classic acceptance-rejection method, where we treat the unconstrained policy as the proposal distribution and derive a modified policy with feasible actions. (ii) To improve the acceptance rate of the proposal distribution, we construct an augmented two-objective Markov decision process (MDP), which include additional self-loop state transitions and a penalty signal for the rejected actions. This augmented MDP incentives the learned policy to stay close to the feasible action sets. Through extensive experiments in both robot control and resource allocation domains, we demonstrate that the proposed framework enjoys faster training progress, better constraint satisfaction, and a lower action inference time simultaneously than the state-of-the-art ACRL methods. We have made the source code publicly available to encourage further research in this direction.

Policy learning under action constraints plays a central role in ensuring safe behaviors in various robot control and resource allocation applications. In this paper, we study a new problem setting termed Action-Constrained Imitation Learning (ACIL), where an action-constrained imitator aims to learn from a demonstrative expert with larger action space. The fundamental challenge of ACIL lies in the unavoidable mismatch of occupancy measure between the expert and the imitator caused by the action constraints. We tackle this mismatch through trajectory alignment and propose DTWIL, which replaces the original expert demonstrations with a surrogate dataset that follows similar state trajectories while adhering to the action constraints. Specifically, we recast trajectory alignment as a planning problem and solve it via Model Predictive Control, which aligns the surrogate trajectories with the expert trajectories based on the Dynamic Time Warping (DTW) distance. Through extensive experiments, we demonstrate that learning from the dataset generated by DTWIL significantly enhances performance across multiple robot control tasks and outperforms various benchmark imitation learning algorithms in terms of sample efficiency.

Human-like agents have long been one of the goals in pursuing artificial intelligence. Although reinforcement learning (RL) has achieved superhuman performance in many domains, relatively little attention has been focused on designing human-like RL agents. As a result, many reward-driven RL agents often exhibit unnatural behaviors compared to humans, raising concerns for both interpretability and trustworthiness. To achieve human-like behavior in RL, this paper first formulates human-likeness as trajectory optimization, where the objective is to find an action sequence that closely aligns with human behavior while also maximizing rewards, and adapts the classic receding-horizon control to human-like learning as a tractable and efficient implementation. To achieve this, we introduce Macro Action Quantization (MAQ), a human-like RL framework that distills human demonstrations into macro actions via Vector-Quantized VAE. Experiments on D4RL Adroit benchmarks show that MAQ significantly improves human-likeness, increasing trajectory similarity scores, and achieving the highest human-likeness rankings among all RL agents in the human evaluation study. Our results also demonstrate that MAQ can be easily integrated into various off-the-shelf RL algorithms, opening a promising direction for learning human-like RL agents. Our code is available at https://rlg.iis.sinica.edu.tw/papers/MAQ.

Proximal Policy Optimization algorithm employing a clipped surrogate objective (PPO-Clip) is a prominent exemplar of the policy optimization methods. However, despite its remarkable empirical success, PPO-Clip lacks theoretical substantiation to date. In this paper, we contribute to the field by establishing the first global convergence results of a PPO-Clip variant in both tabular and neural function approximation settings. Our findings highlight the O(1/√T ) min-iterate convergence rate specifically in the context of neural function approximation. We tackle the inherent challenges in analyzing PPO-Clip through three central concepts: (i) We introduce a generalized version of the PPO-Clip objective, illuminated by its connection with the hinge loss. (ii) Employing entropic mirror descent, we establish asymptotic convergence for tabular PPO-Clip with direct policy parameterization. (iii) Inspired by the tabular analysis, we streamline convergence analysis by introducing a two-step policy improvement approach. This decouples policy search from complex neural policy parameterization using a regression-based update scheme. Furthermore, we gain deeper insights into the efficacy of PPO-Clip by interpreting these generalized objectives. Our theoretical findings also mark the first characterization of the influence of the clipping mechanism on PPO-Clip convergence. Importantly, the clipping range affects only the pre-constant of the convergence rate.

Various acceleration approaches for Policy Gradient (PG) have been analyzed within the realm of Reinforcement Learning (RL). However, the theoretical understanding of the widely used momentum-based acceleration method on PG remains largely open. In response to this gap, we adapt the celebrated Nesterov's accelerated gradient (NAG) method to policy optimization in RL, termed *Accelerated Policy Gradient* (APG). To demonstrate the potential of APG in achieving fast convergence, we formally prove that with the true gradient and under the softmax policy parametrization, APG converges to an optimal policy at rates: (i) $\tilde{O}(1/t^2)$ with nearly constant step sizes; (ii) $O(e^{-ct})$ with time-varying step sizes. To the best of our knowledge, this is the first characterization of the convergence rates of NAG in the context of RL. Notably, our analysis relies on one interesting finding: Regardless of the parameter initialization, APG ends up entering a locally nearly-concave regime, where APG can significantly benefit from the momentum, within finite iterations. Through numerical validation and experiments on the Atari 2600 benchmarks, we confirm that APG exhibits a $\tilde{O}(1/t^2)$ rate with nearly constant step sizes and a linear convergence rate with time-varying step sizes, significantly improving convergence over the standard PG.

In offline reinforcement learning (RL), updating the value function with the discrete-time Bellman Equation often encounters challenges due to the limited scope of available data. This limitation stems from the Bellman Equation, which cannot accurately predict the value of unvisited states. To address this issue, we have introduced an innovative solution that bridges the continuous- and discrete-time RL methods, capitalizing on their advantages. Our method uses a discrete-time RL algorithm to derive the value function from a dataset while ensuring that the function’s first derivative aligns with the local characteristics of states and actions, as defined by the Hamilton-Jacobi-Bellman equation in continuous RL. We provide practical algorithms for both deterministic policy gradient methods and stochastic policy gradient methods. Experiments on the D4RL dataset show that incorporating the first-order information significantly improves policy performance for offline RL problems.

Imitation learning aims to learn a policy from observing expert demonstrations without access to reward signals from environments. Generative adversarial imitation learning (GAIL) formulates imitation learning as adversarial learning, employing a generator policy learning to imitate expert behaviors and discriminator learning to distinguish the expert demonstrations from agent trajectories. Despite its encouraging results, GAIL training is often brittle and unstable. Inspired by the recent dominance of diffusion models in generative modeling, we propose Diffusion-Reward Adversarial Imitation Learning (DRAIL), which integrates a diffusion model into GAIL, aiming to yield more robust and smoother rewards for policy learning. Specifically, we propose a diffusion discriminative classifier to construct an enhanced discriminator, and design diffusion rewards based on the classifier’s output for policy learning. Extensive experiments are conducted in navigation, manipulation, and locomotion, verifying DRAIL’s effectiveness compared to prior imitation learning methods. Moreover, additional experimental results demonstrate the generalizability and data efficiency of DRAIL. Visualized learned reward functions of GAIL and DRAIL suggest that DRAIL can produce more robust and smoother rewards. Project page: https://nturobotlearninglab.github.io/DRAIL/

Reward-biased maximum likelihood estimation (RBMLE) is a classic principle in the adaptive control literature for tackling explore-exploit trade-offs. This paper studies the neural contextual bandit problem from a distributional perspective and proposes NeuralRBMLE, which leverages the likelihood of surrogate parametric distributions to learn the unknown reward distributions and thereafter adapts the RBMLE principle to achieve efficient exploration by properly adding a reward-bias term. NeuralRBMLE leverages the representation power of neural networks and directly encodes exploratory behavior in the parameter space, without constructing confidence intervals of the estimated rewards. We propose two variants of NeuralRBMLE algorithms: The first variant directly obtains the RBMLE estimator by gradient ascent, and the second variant simplifies RBMLE to a simple index policy through an approximation. We show that both algorithms achieve order-optimality. Through extensive experiments, we demonstrate that the NeuralRBMLE algorithms achieve comparable or better empirical regrets than the state-of-the-art methods on real-world datasets with non-linear reward functions.

We revisit the domain of off-policy policy optimization in RL from the perspective of coordinate ascent. One commonly-used approach is to leverage the off-policy policy gradient to optimize a surrogate objective – the total discounted in expectation return of the target policy with respect to the state distribution of the behavior policy. However, this approach has been shown to suffer from the distribution mismatch issue, and therefore significant efforts are needed for correcting this mismatch either via state distribution correction or a counterfactual method. In this paper, we rethink off-policy learning via Coordinate Ascent Policy Optimization (CAPO), an off-policy actor-critic algorithm that decouples policy improvement from the state distribution of the behavior policy without using the policy gradient. This design obviates the need for distribution correction or importance sampling in the policy improvement step of off-policy policy gradient. We establish the global convergence of CAPO with general coordinate selection and then further quantify the convergence rates of several instances of CAPO with popular coordinate selection rules, including the cyclic and the randomized variants of CAPO. We then extend CAPO to neural policies for a more practical implementation. Through experiments, we demonstrate that CAPO provides a competitive approach to RL in practice.

Domain randomization (DR) is widely used in reinforcement learning (RL) to bridge the gap between simulation and reality by maximizing its average returns under the perturbation of environmental parameters. However, even the most complex simulators cannot capture all details in reality due to finite domain parameters and simplified physical models. Additionally, the existing methods often assume that the distribution of domain parameters belongs to a specific family of probability functions, such as normal distributions, which may not be correct. To overcome these limitations, we propose a new approach to DR by rethinking it from the perspective of adversarial state perturbation, without the need for reconfiguring the simulator or relying on prior knowledge about the environment. We also address the issue of over-conservatism that can occur when perturbing agents to the worst states during training by introducing a Relaxed State-Adversarial Algorithm that simultaneously maximizes the average-case and worst-case returns. We evaluate our method by comparing it to state-of-the-art methods, providing experimental results and theoretical proofs to verify its effectiveness. Our source code and appendix are available at https://github.com/sophialien/RAPPO.

Traffic splitting is to distribute traffic over multiple paths so as to better utilize link bandwidth and avoid potential link congestion. Many traffic splitting schemes have been proposed for datacenter networks, which provide different levels of splitting granularity. This paper proposes a packet-level traffic splitting scheme, which provides the most fine-grained link utilization as it lays out a forwarding path for each individual packet. The proposed scheme mitigates possible packet reordering problem by equalizing the build-ups of packet queues in all switches as the path layout utilizes links in proportional to the associated exploitable link bandwidth. The result is a packet dispatching rule designated to each switch. We implemented the rules using P4 (Programming Protocol-Independent Packet Processors) switches and conducted experiments to measure the performance. Experimental results show that the proposed scheme provides higher goodput compared with another packet-level traffic splitting scheme which dispatches packets randomly among all links.

This paper presents a reinforcement learning (RL) framework that utilizes Frank-Wolfe policy optimization to solve Coding- Tree-Unit (CTU) bit allocation for Region-of-Interest (ROI) intra-frame coding. Most previous RL-based methods employ the single-critic design, where the rewards for distortion minimization and rate regularization are weighted by an empirically chosen hyper-parameter. Recently, the dual-critic design is proposed to update the actor by alternating the rate and distortion critics. However, its convergence is not guaranteed. To address these issues, we introduce Neural Frank-Wolfe Policy Op-timization (NFWPO) in formulating the CTU-level bit allocation as an action-constrained RL problem. In this new framework, we exploit a rate critic to predict a feasible set of actions. With this feasible set, a distortion critic is invoked to update the actor to maximize the ROI-weighted image quality subject to a rate constraint. Experimental results produced with x265 confirm the superiority of the proposed method to the other baselines.

Modifying the reward-biased maximum likelihood method originally proposed in the adaptive control literature, we propose novel learning algorithms to handle the explore-exploit trade-off in linear bandits problems as well as generalized linear bandits problems. We develop novel index policies that we prove achieve order-optimality, and show that they achieve empirical performance competitive with the state-of-the-art benchmark methods in extensive experiments. The new policies achieve this with low computation time per pull for linear bandits, and thereby resulting in both favorable regret as well as computational efficiency.

This paper studies a remote sensing system where multiple wireless sensors generate possibly noisy information updates of various surveillance fields and delivering these updates to a control center over a wireless network. The control center needs a sufficient number of recently generated information updates to have an accurate estimate of the current system status, which is critical for the control center to make appropriate control decisions. The goal of this work is then to design the optimal policy for scheduling the transmissions of information updates. Through Brownian approximation, we demonstrate that the control center's ability to make accurate real-time estimates depends on the averages and temporal variances of the delivery processes. We then formulate a constrained optimization problem to find the optimal means and variances. We also develop a simple online scheduling policy that employs the optimal means and variances to achieve the optimal system-wide performance. Simulation results show that our scheduling policy enjoys fast convergence speed and better performance when compared to other state-of-the-art policies.

Bayesian optimization (BO) conventionally relies on handcrafted acquisition functions (AFs) to sequentially determine the sample points. However, it has been widely observed in practice that the best-performing AF in terms of regret can vary significantly under different types of black-box functions. It has remained a challenge to design one AF that can attain the best performance over a wide variety of black-box functions. This paper aims to attack this challenge through the perspective of reinforced few-shot AF learning (FSAF). Specifically, we first connect the notion of AFs with Q-functions and view a deep Q-network (DQN) as a surrogate differentiable AF. While it serves as a natural idea to combine DQN and an existing few-shot learning method, we identify that such a direct combination does not perform well due to severe overfitting, which is particularly critical in BO due to the need of a versatile sampling policy. To address this, we present a Bayesian variant of DQN with the following three features: (i) It learns a distribution of Q-networks as AFs based on the Kullback-Leibler regularization framework. This inherently provides the uncertainty required in sampling for BO and mitigates overfitting. (ii) For the prior of the Bayesian DQN, we propose to use a demo policy induced by an off-the-shelf AF for better training stability. (iii) On the meta-level, we leverage the meta-loss of Bayesian model-agnostic meta-learning, which serves as a natural companion to the proposed FSAF. Moreover, with the proper design of the Q-networks, FSAF is general-purpose in that it is agnostic to the dimension and the cardinality of the input domain. Through extensive experiments, we demonstrate that the FSAF achieves comparable or better regrets than the state-of-the-art benchmarks on a wide variety of synthetic and real-world test functions.

Whittle index policy is a powerful tool to obtain asymptotically optimal solutions for the notoriously intractable problem of restless bandits. However, finding the Whittle indices remains a difficult problem for many practical restless bandits with convoluted transition kernels. This paper proposes NeurWIN, a neural Whittle index network that seeks to learn the Whittle indices for any restless bandits by leveraging mathematical properties of the Whittle indices. We show that a neural network that produces the Whittle index is also one that produces the optimal control for a set of Markov decision problems. This property motivates using deep reinforcement learning for the training of NeurWIN. We demonstrate the utility of NeurWIN by evaluating its performance for three recently studied restless bandit problems.Our experiment results show that the performance of NeurWIN is significantly better than other RL algorithms.

Action-constrained reinforcement learning (RL) is a widely-used approach in various real-world applications, such as scheduling in networked systems with resource constraints and control of a robot with kinematic constraints. While the existing projection-based approaches ensure zero constraint violation, they could suffer from the zero-gradient problem due to the tight coupling of the policy gradient and the projection, which results in sample-inefficient training and slow convergence. To tackle this issue, we propose a learning algorithm that decouples the action constraints from the policy parameter update by leveraging state-wise Frank-Wolfe and a regression-based policy update scheme. Moreover, we show that the proposed algorithm enjoys convergence and policy improvement properties in the tabular case as well as generalizes the popular DDPG algorithm for action-constrained RL in the general case. Through experiments, we demonstrate that the proposed algorithm significantly outperforms the benchmark methods on a variety of control tasks.

Inspired by the Reward-Biased Maximum Likelihood Estimate method of adaptive control, we propose RBMLE – a novel family of learning algorithms for stochastic multi-armed bandits (SMABs). For a broad range of SMABs including both the parametric Exponential Family as well as the non-parametric sub-Gaussian/Exponential family, we show that RBMLE yields an index policy. To choose the bias-growth rate $\alpha(t)$ in RBMLE, we reveal the nontrivial interplay between $\alpha(t)$ and the regret bound that generally applies in both the Exponential Family as well as the sub-Gaussian/Exponential family bandits. To quantify the finite-time performance, we prove that RBMLE attains order-optimality by adaptively estimating the unknown constants in the expression of $\alpha(t)$ for Gaussian and sub-Gaussian bandits. Extensive experiments demonstrate that the proposed RBMLE achieves empirical regret performance competitive with the state-of-the-art methods, while being more computationally efficient and scalable in comparison to the best-performing ones among them.

This paper presents a Brownian-approximation framework to optimize the quality of experience (QoE) for real-time video streaming in wireless networks. In real-time video streaming, one major challenge is to tackle the natural tension between the two most critical QoE metrics: playback latency and video interruption. To study this trade-off, we first propose an analytical model that precisely captures all aspects of the playback process of a real-time video stream, including playback latency, video interruptions, and packet dropping. Built on this model, we show that the playback process of a real-time video can be approximated by a two-sided reflected Brownian motion. Through such Brownian approximation, we are able to study the fundamental limits of the two QoE metrics and characterize a necessary and sufficient condition for a set of QoE performance requirements to be feasible. We propose a scheduling policy that satisfies any feasible set of QoE performance requirements and then obtain simple rules on the trade-off between playback latency and the video interrupt rates, in both heavy-traffic and under-loaded regimes. Finally, simulation results verify the accuracy of the proposed approximation and show that the proposed policy outperforms other popular baseline policies.

Sequential decision making for lifetime maximization is a critical problem in many real-world applications, such as medical treatment and portfolio selection. In these applications, a “reneging” phenomenon, where participants may disengage from future interactions after observing an unsatisfiable outcome, is rather prevalent. To address the above issue, this paper proposes a model of heteroscedastic linear bandits with reneging, which allows each participant to have a distinct “satisfaction level," with any interaction outcome falling short of that level resulting in that participant reneging. Moreover, it allows the variance of the outcome to be context-dependent. Based on this model, we develop a UCB-type policy, namely HR-UCB, and prove that it achieves $\mathcal{O}\big(\sqrt{{T}(\log({T}))^{3}}\big)$ regret. Finally, we validate the performance of HR-UCB via simulations.

This paper studies coverage enhancement of LTE Cat-M1 devices from an experimental perspective. We first conduct link-level lab experiments on Cat-M1 modules using an RF test system and show the coverage gain as well as link throughput versus different repetition numbers. From these measurement results, we observe significant coverage gain resulting from the repetition mechanism, at the expense of a lower throughput. To mitigate the reduction in throughput, we further propose adaptive repetition algorithms to achieve higher throughput while maintaining the same coverage. The simulation results show that the proposed algorithms achieve significantly higher throughput than the fixed-repetition approach.

This paper proposes a feasibility-optimal decentralized algorithm for real-time wireless ad hoc networks, where a strict deadline is imposed for each packet. While centralized scheduling algorithms provide provably optimal theoretical guarantees, they may not be practical in many settings, such as industrial networked control systems. Therefore, it is of great importance to design an algorithm that achieves feasibility optimality in a decentralized manner. To design a decentralized algorithm, we leverage two widely-used functions of wireless devices: carrier sensing and backoff timers. Different from the conventional approach, the proposed algorithm utilizes a collision-free backoff scheme to enforce the transmission priority of different links. This design obviates the capacity loss due to collision with quantifiably small backoff overhead. The algorithm is fully decentralized in the sense that every link only needs to know its own priority, and links contend for priorities only through carrier sensing. We prove that the proposed algorithm is feasibility-optimal. NS-3 simulation results show that the proposed algorithm indeed performs as well as the feasibility-optimal centralized algorithm. Moreover, the results also show that the proposed algorithm converges to optimality very fast.

An increasing number of applications that will be supported by next generation wireless networks require packets to arrive before a certain deadline for the system to have the desired performance. While many time-sensitive scheduling protocols have been proposed, few have been experimentally evaluated to establish realistic performance. Furthermore, some of these protocols involve high complexity algorithms that need to be performed on a per-packet basis. Experimental evaluation of these protocols requires a flexible platform that is readily capable of implementing and experimenting with these protocols.

Ultra-low per-packet latency has become an essential system requirement as well as a critical challenge for wireless networks. While there is a rich literature on real-time wireless scheduling, it is still unclear what the minimum achievable latency is and what level of throughput can be obtained in practice. This demo presents PULS, a processor-supported software-defined wireless testbed that supports ultra-low-latency scheduling protocols. We will demonstrate that PULS provides strict per-packet latency guarantees as low as 1 millisecond with realistic throughput for wireless networks.

The emerging connected-vehicle technology provides a new dimension for developing more intelligent traffic control algorithms for signalized intersections. An important challenge for scheduling in networked transportation systems is the switchover delay caused by the guard time before any traffic signal change. The switch-over delay can result in significant loss of system capacity and hence needs to be accommodated in the scheduling design. To tackle this challenge, we propose a distributed online scheduling policy that extends the well-known Max-Pressure policy to address switch-over delay by introducing a bias factor favoring the current schedule. We prove that the proposed policy is throughput-optimal with switch-over delay. Furthermore, the proposed policy remains optimal when there are both connected signalized intersections and conventional fixed-time ones in the system. With connected-vehicle technology, the proposed policy can be easily incorporated into the current transportation systems without additional infrastructure. Through extensive simulation in VISSIM, we show that our policy indeed outperforms the existing popular policies.

This paper proposes online scheduling policies to optimize quality of experience (QoE) for video-on-demand applications in wireless networks. We consider wireless systems where an access point (AP) transmits video content to clients over fading channels. The QoE of each flow is measured by its duration of video playback interruption. We are specifically interested in systems operating in the heavy-traffic regime. We first consider a special case of ON-OFF channels and establish a scheduling policy that achieves every point in the capacity region under heavy-traffic conditions. This policy is then extended for more general fading channels, and we prove that it remains optimal under some mild conditions. We then formulate a network utility maximization problem based on the QoE of each flow. We demonstrate that our policies achieve the optimal overall utility when their parameters are chosen properly. Finally, we compare our policies against three popular policies. Simulation results validate that the proposed policy indeed outperforms existing policies.

Video streaming is anticipated to dominate wireless traffic in the near future. We study wireless systems where an access point delivers video streams to multiple clients over unreliable wireless channels. The performance of each client is measured by the amount of time that its video playback halts due to buffer underflow, which has been shown to have the most impact on client's perceived quality of experience (QoE).

This demo presents WiMAC, a general-purpose wireless testbed for researchers to quickly prototype a wide variety of real-time MAC protocols for wireless networks. As the interface between the link layer and the physical layer, MAC protocols are often tightly coupled with the underlying physical layer, and need to have extremely small latencies. Implementing a new MAC requires a long time. In fact, very few MACs have ever been implemented, even though dozens of new MAC protocols have been proposed. To enable quick prototyping, we employ the mechanism vs. policy separation to decompose the functionality in the MAC layer and the PHY layer. Built on the separation framework, WiMAC achieves the independence of the software from the hardware, offering a high degree of function reuse and design flexibility. Hence, our platform not only supports easy cross-layer design but also allows protocol changes on the fly. Following the 802.11-like reference design, we demonstrate that deploying a new MAC protocol is quick and simple on the proposed platform through the implementation of the CSMA/CA and CHAIN protocols.