Recently, Multimodal Large Language Models (MLLMs) have demonstrated their immense potential in computer-aided diagnosis and decision-making. In the context of robotic-assisted surgery, MLLMs can serve as effective tools for surgical training and guidance. However, there is still a deficiency of MLLMs specialized for surgical scene understanding in endoscopic procedures. To this end, we present EndoChat, an MLLM tailored to address various dialogue paradigms and subtasks in understanding endoscopic procedures. To train our EndoChat, we construct the Surg-396K dataset through a novel pipeline that systematically extracts surgical information and generates structured annotations based on large-scale endoscopic surgery datasets. Furthermore, we introduce a multi-scale visual token interaction mechanism and a visual contrast-based reasoning mechanism to enhance the model's representation learning and reasoning capabilities. Our model achieves state-of-the-art performance across five dialogue paradigms and seven surgical scene understanding tasks. Additionally, we conduct evaluations with professional surgeons, who provide positive feedback on the majority of conversation cases generated by EndoChat. Overall, these results demonstrate that EndoChat has the potential to advance training and automation in robotic-assisted surgery. Our dataset and model are publicly available at https://github.com/gkw0010/EndoChat.

With the increasing integration of renewable energy into power systems, effective long-term planning of hybrid energy resources—particularly those involving storage—is becoming essential to ensure reliability and cost-efficiency. This paper presents a planning framework for hybrid systems that incorporate wind, solar, diesel generation, and battery storage, with explicit consideration of battery state-of-health degradation and automated supplementation strategies. A lifecycle optimization model is developed to jointly determine battery installation timing, sizing, and technology selection, while accounting for degradation dynamics over time. To manage the high dimensionality introduced by long planning horizons and hourly resolution, the framework employs techniques such as battery variable aggregation, bounds tightening, and hyperparameter optimization. These methods enable full-system optimization within 1,000–2,000 seconds. The study examines two system configurations: wind-solar-battery and wind-solar-diesel-battery. Compared to baseline models that use heuristic battery replacement strategies, the proposed model achieves reductions in Levelized Cost of Energy by 5.1% and 4.3%, respectively. When photovoltaic degradation (0.1%–0.2% per year) is incorporated, the cost-saving benefits of the proposed model become even more significant, without increasing computational complexity. Overall, the results underscore the importance of degradation-aware, full-horizon optimization for achieving cost-effective and realistic energy system planning, especially in large-scale renewable deployments with integrated storage.

Traffic Engineering (TE) represents one of the most essential tools in modern telecommunication networks. The rapid growth of exchanged traffic has required tackling a known NP-hard problem called the Multi-Commodity Flow problem (MCF). Many studies in the literature have already considered different variants of this problem. In this paper, we propose a new way to use a double Dantzig-Wolfe decomposition formulation to improve the quality of the linear relaxation. We apply our method on the classical multi-commodity flow problem where the throughput acceptance is first maximized and then the routing cost is minimized. We provide a computational experiment and conduct an in-depth analysis of the algorithm based on realistic instances.

In this paper, we propose a Bi-layer Prediction-based Reduction Branch (BP-RB) framework to speed up the process of finding a high-quality feasible solution for Mixed Integer Programming (MIP) problems. A graph convolutional network (GCN) is employed to predict binary variables' values. After that, a subset of binary variables is fixed to the predicted value by a greedy method conditioned on the predicted probabilities. By exploring the logical consequences, a learning-based problem reduction method is proposed, significantly reducing the variable and constraint sizes. With the reductive sub- MIP problem, the second layer GCN framework is employed to update the prediction for the remaining binary variables' values and to determine the selection of variables which are then used for branching to generate the Branch and Bound (B&B) tree. Numerical examples show that our BP-RB framework speeds up the primal heuristic and finds the feasible solution with high quality.

Despite that cloud virtual reality (VR) is the most promising service in 5G networks, a reasonable traffic model for it is still unknown. Based on statistics of real cloud VR traces, we justify that the Markovian arrival process (MAP) can well characterize the inter-arrival times (IATs) among packets. Moreover, we discuss possible methods to efficiently estimate parameters for flow aggregation. Since MAPs are analytically tractable, we quantify the performance of delivering cloud VR traffic from a queueing theory perspective. Aiming at supporting quantile metrics, we provide distributions of queue-length and latency for an arbitrary packet. The accuracy of estimated latency is validated by comparing with measured latency on an industrial 5G platform. The MAP-based traffic model will potentially serve as an input for performance evaluation and network planning for assuring high requirements of user experience.

Predicting block popularity is of crucial importance for data placement in multi-tiered multimedia storage systems. Traditional methods, such as least recently used and exponential smoothing, are commonly employed to predict future block access frequencies and fail to achieve good performance for complex and changing access patterns. Recently, deep neural networks have brought great success to pattern recognition and prediction, which motivates us to introduce deep learning to solve the problem of block popularity prediction. In this paper, we first analyze and verify the temporal and spatial correlations among the multimedia I/O traces. Then, we design a multi-dimension feature to capture such correlations, which serves as the input of the designed deep neural network. A spatial-temporal-sequential neural network (STSNN) and its variants that capture the locality information, time dependency information, and block sequential information are proposed to predict the block popularity. We systematically evaluate our STSNN models against six baseline models from three different categories, i.e., heuristic methods, regression methods and neural network-based methods. Experiment results show that our proposed STSNN models are very promising for predicting block access frequencies under some of Huawei and Microsoft datasets and particularly achieve 2-6 times better performance compared with the baselines in terms of the I/O hit ratio, I/O recall rate and I/O prediction ratio under the Microsoft 64 MB-block dataset.

Cloud providers have started to deploy various FPGA accelerators in their datacenters because the performance of many applications can be significantly improved by implementing their core routines in FPGAs. In conventional datacenters with FPGA accelerated servers, if a tenant wants to use FPGA accelerators, it requests for a VM instance residing in a server equipped with an FPGA accelerator. This paradigm to integrate FPGA into Cloud leads to poor resource sharing of the precious FPGA resources. In this paper, we propose FPGAPooling, an FPAG-enabled Cloud system where all FPGA accelerators are managed as a single resource pool and shared among all VMs. For a VM, instead of requesting the Cloud to run the VM on an FPGA accelerated server, at run time, when a VM needs to use FPGA acceleration, it requests an FPGA accelerator from the pool. After the VM finishes using the FPGA accelerator, it releases the FPGA accelerator back to the pool. We design a centralized scheduler to handle acceleration requests from VMs and assign each request to an idle FPGA accelerator at run time; We implemented a system prototype on IBM's OpenPower Cloud system. The key challenging of FPGAPooling is scheduling. We designed and implemented a group of scheduling algorithms for the FPGAPooling system. With extensive evaluations on both a small testbed and a large-scale simulation, we found that our algorithms can improve the average and tail job completion time by up to 7 and 4 times, respectively.

In this paper, we consider a MIMO cyber physical system (CPS) in which a sensor amplifies and forwards the observed MIMO plant state to a remote controller via a MIMO fading channel. We focus on the MIMO amplify-and-forward (AF) precoding design at the sensor to minimize a weighted average state estimation error at the remote controller subject to an average communication power gain constraint of the sensor. The MIMO AF precoding design is formulated as an infinite horizon average cost Markov decision process (MDP). To deal with the curse of dimensionality associated with the MDP, we propose a novel continuous-time perturbation approach and derive an asymptotically optimal closed-form priority function for the MDP. Based on this, we derive a closed-form first-order optimal dynamic MIMO AF precoding solution, and the solution has an event-driven control structure. Specifically, the sensor activates the strongest eigenchannel to deliver a dynamically weighted combination of the plant states to the controller when the accumulated state estimation error exceeds a dynamic threshold. We further establish technical conditions for ensuring the stability of the MIMO CPS.

In memory computing and high-end distributed storage demand low latency, high throughput, and zero data loss simultaneously from datacenter networks. Existing reactive congestion control approaches cannot both minimize queuing latency and ensure zero data loss. A token-oriented proactive approach can achieve them together by controlling congestion even before sending data packets. However, state-of-the-art token-oriented approaches only strive to optimize network-level metrics: maximizing throughput while achieving flow-level fairness. This article answers the question of how to support objective-aware traffic scheduling in token-oriented approaches. The novelty of Token-Oriented in-network Prioritization (TOP) is that it prioritizes tokens instead of data packets. We make three contributions. Via simulations over a hypothetical TOP system, our first contribution is demonstrating the potential performance gain that can be brought by TOP. Second, we investigate the applicability of TOP. Although the overhead of enabling necessary TOP features in switches is trivial, we find that mainstream commodity datacenter switches do not support them. We hence propose a readily-deployable remedy to achieve in-network prioritization by pushing both switch and end-host hardware capacity to an extreme end. Lastly, we implement a running TOP system with Linux hosts and commodity switches, and evaluate TOP in testbeds and with large-scale simulations for various scenarios.

In data center networks, flows with different objectives coexist and compete for limited resources. From the application-level perspective, it is hard to satisfy the demands of different flows without effective resource planning. To address the bandwidth allocation problem under multi-objective scenarios, we study a multi-objective network utility maximization (NUM) problem and propose our practical bandwidth allocation framework HOMMO, which consists of an upper layer algorithm (ULA) and a lower layer algorithm (LLA). This hierarchical design helps to strike a balance between accuracy and efficiency. Implemented in network switches, ULA is an online learning-based scheme that allocates bandwidth for aggregated flows with different performances objectives. Taking the outputs from ULA as capacity constraints, LLA acts as a fast scheduling method at packet-level. We extend NUMFabric by implementing the design of isolate queues and corresponding dequeue strategy in switches and make it a feasible solution of the LLA. Therefore, HOMMO achieves satisfactory isolation across flows with different objectives, which is equivalent to provide a network slicing solution. To evaluate the proposed framework, we implement it in ns-3 and verify the performance under various scenarios. The simulation results show that HOMMO not only quickly converges to a near-optimal solution of the multi-objective NUM problem but also guarantees a Pareto-optimal solution. Moreover, it outperforms the well-known transport protocol (i.e., DCTCP) with 2x increase on average bandwidth utilization and 1.96x improvement in global network utility at the aggregated flow level.

Providing backhaul connectivity for macro and pico base stations (BSs) constitutes a significant share of infrastructure costs in future heterogeneous networks (HetNets). To address this issue, the emerging idea of flexible backhaul is proposed. Under this architecture, not all the pico BSs are connected to the backhaul, resulting in a significant reduction in the infrastructure costs. In this regard, pico BSs without backhaul connectivity need to communicate with their nearby BSs in order to have indirect accessibility to the backhaul. This makes the radio resource management (RRM) in such networks more complex and challenging. In this paper, we address the problem of cross-layer RRM in HetNets with flexible backhaul. We formulate this problem as a two-timescale non-convex stochastic optimization, which jointly optimizes flow control, routing, interference mitigation, and link scheduling in order to maximize a generic network utility. By exploiting a hidden convexity of this non-convex problem, we propose an iterative algorithm which converges to the global optimal solution. The proposed algorithm benefits from low complexity and low signaling, which makes it scalable. Moreover, due to the proposed two-timescale design, it is robust to the backhaul signaling latency as well. Simulation results demonstrate the significant performance gain of the proposed solution over various baselines.

In datacenter networks, flows can have different performance objectives. We use a tenant-objective division to denote all flows of a tenant that share the same objective. Bandwidth allocation in datacenters should support not only performance isolation among divisions but also objective-oriented scheduling among flows within the same division. This paper studies the Multi-Tenant Multi-Objective (MT-MO) bandwidth allocation problem. To our best knowledge, no existing practical work support performance isolation and objective scheduling simultaneously. We propose Stacked Congestion Control (SCC), a distributed host-based bandwidth allocation design, where an underlay congestion control (UCC) layer handles contention among divisions, and a private congestion control (PCC) layer for each division optimizes its performance objective. Via the tenant-objective tunnel abstraction, SCC achieves weighted bandwidth sharing for each division in a distributed and transparent way. By adding a rate-limiting send queue in the ingress of each tunnel, mechanisms between performance isolation and objective scheduling are completely decoupled. We evaluate SCC both on a small-scale testbed and with large-scale NS-2 simulations. Compared to the direct coexistence cases, SCC reduces latency by up to 40% for Latency-Sensitive flows, deadline miss ratio by up to 3.2× for Deadline-Sensitive flows, and average flow-completion-time by up to 53% for Completion-Sensitive flows.

In this paper, we consider the problem of cross-layer radio resource management in heterogeneous networks (HetNets) with flexible backhaul, which aims at minimizing the total transmit power of base stations (BSs) while guaranteeing the average end-to-end data rate requirement of each data flow. We formulate the problem as a two-time scale stochastic optimisation, where the long-timescale control variables are flow control, routing control and interference coordination, while the short-timescale control variable is instantaneous beamforming within each cell. Using a stochastic cutting plane (SCP) method, we propose a cross-layer queue-state information (QSI) aware radio resource management (RRM) solution in which the long-term controls are updated centrally at radio resource management server (RRMS) without needing to know the global statistical information of the network, and the short-term control variables are updated locally at each BS with only the local instantaneous QSI and channel-state information (CSI) available at each cell. Simulation results show the significant performance gain of our proposed algorithm compared to various baselines.

In this paper, we focus on the problem of hierarchical cross-layer dynamic resource allocation for heterogeneous networks with flexible backhaul. We formulate the radio resource management problem as a two timescale non-convex stochastic optimisation problem which jointly optimizes flow control, routing control, interference mitigation and link scheduling in order to maximize a generic network utility. We propose an iterative hierarchical control structure where the long-term controls are adaptive to large scale fading and the short-term control is adaptive to the local CSI within a pico or macro BS, to find the optimal solution. The proposed solution benefits from low complexity and requires low signalling and message passing among different nodes, which makes it scalable. Moreover, due to the proposed two-timescale hierarchical design, it is robust to the backhaul signalling latency as well. Simulation results demonstrate the significant performance gain of the proposed solution over various baselines.

One of the main advantages of utilizing flexible backhaul in future HetNets is that it can provide better user experience through flexible resource allocation. For this purpose, it is important to provide the required quality of service (QoS) in such networks. In this paper, we consider the problem of cross-layer radio resource management in HetNets with flexible backhaul which guarantees minimum total transmit power of BSs as well as the average end-to-end data rate requirement of each data flow. The problem is formulated as a two-timescale stochastic optimisation, where the long-timescale control variables are flow control, routing control and interference mitigation, while the short-timescale control variable is instantaneous beamforming within each cell. Using stochastic cutting plane, we propose an online cross-layer hierarchical algorithm in which the long-term controls are updated centrally at radio resource management server (RRMS) without needing to know the global statistical information of the network, and the short-term control variables are updated locally at each BS with only the local instantaneous CSI available at each cell. The proposed algorithm has low complexity and signalling overhead. Moreover, simulation results show the significant gains of our proposed algorithm over various baselines.

In this paper, we study the grid power-delay tradeoff in a point-to-point energy harvesting wireless communication system with finite energy storage capacity serving delay-sensitive applications. This communication system is powered by both grid and renewable power sources. First, we consider the average grid power consumption minimization subject to the data queue stability and renewable energy availability constraints. By exploring the optimality property and using the theory of random walks, we transform the grid power minimization problem to an asymptotically equivalent problem. Using the Lyapunov drift approach, we obtain an online dynamic power control policy to solve the asymptotically equivalent problem. Then, we introduce a novel analysis framework to study the grid power-delay tradeoff relationship of the online power control in the small delay regime. Specifically, using continuous-time approximation, dynamic programming and sample-path approach, we obtain bounds on the average delay and grid power consumption, which are asymptotically tight in the small delay regime. Based upon the derived closed-form expressions, we quantify the impacts of energy storage capacity and some other system parameters on the grid power-delay tradeoff.

In this paper, we consider the dynamic power control for delay-aware D2D communications. The stochastic optimization problem is formulated as an infinite horizon average cost Markov decision process. To deal with the curse of dimensionality, we utilize the interference filtering property of the CSMA-like MAC protocol and derive a closed-form approximate priority function and the associated error bound using perturbation analysis. Based on the closed-form approximate priority function, we propose a low-complexity power control algorithm solving the per-stage optimization problem. The proposed solution is further shown to be asymptotically optimal for a sufficiently large carrier sensing distance. Finally, the proposed power control scheme is compared with various baselines through simulations, and it is shown that significant performance gain can be achieved.

In this paper, we consider queue-aware beamforming control for video streaming applications in multiantenna interference network. Using heavy traffic approach, we derive the diffusion limit for the discrete time queuing system and obtain an ergodic excess beamforming control problem for the diffusion limit system. The ergodic control problem is to minimize the average power costs of the access points subject to the constraints on the playback interruption costs and buffer overflow costs of the mobile users. To deal with the queue coupling challenge, we utilize the weak interference coupling property in the network. Using the theories of calculus and perturbation techniques, we derive a closed-form approximate priority function of the optimality equation and the associated error bound. Based on this approximation, we propose a low complexity queue-aware beamforming control algorithm, which is asymptotically optimal for sufficiently small cross-channel path gain. Finally, the proposed scheme is compared with various baselines through simulations and it is shown that significant performance gain can be achieved.

In this paper, we consider an application-driven beamforming control for video streaming in multi-antenna interference networks. The optimization objectives for video streaming are the application level performance metrics, namely the playback interruption cost and buffer overflow cost. Using heavy traffic approximation technique, we first derive the diffusion limit for the discrete time queueing system. Based on the diffusion limit, we then formulate an infinite horizon ergodic control problem to minimize the average power of the base stations subject to the constraints on the playback interruption costs and buffer overflow costs of the mobile users. To deal with the queue coupling challenge, we utilize the weak interference coupling property and derive a closed-form approximate value function of the associated optimality equation using perturbation analysis. Based on the closed- form approximate value function, we propose a low complexity queue-aware beamforming control algorithm, which is asymptotically optimal for sufficiently small cross-channel path gain. Finally, the proposed scheme is compared with various baselines through simulations and it is shown that significant performance gain can be achieved.

Spectrum sensing is adopted to detect the presence of primary users in cognitive radio networks. Considering the sensing overhead, it is not always a good choice that the secondary user senses the channel all the time. In this paper, we propose an opportunistic spectrum sensing decision method according to the primary user's traffic. First, the traffic of primary user is observed and estimated. For estimating the parameters of the primary traffic, the Maximum Likelihood estimation is adopted and the confidence interval is calculated based on finite observed samples. Then, according to the estimated primary traffic information, a decision criterion is proposed for determining whether to sense the channel. The simulation results show that the performance of the proposed opportunistic sensing scheme is much better than that when sensing at every timeslots and very close to that in the ideal case in which the primary traffic information is obtained perfectly.