Congestion Information Research Articles

Layout Congestion Prediction Based on Regression-ViT

To accelerate the back-end design flow of integrated circuit (IC), numerous studies have made exploratory advancements in machine learning (ML) for electronic design automation (EDA). However, most research works are limited to deep learning (DL) models predominantly based on convolutional neural networks, and the models often suffer from poor generalization due to the scarcity of data. In this study, we propose the Double generative adversarial networks (D-GAN) model to enrich the dataset and propose the Regression Vision Transformer (R-ViT) model to predict layout congestion information. Compared to the baseline model, experimental results show improvements of 3.03% and 2.64% in Receiver Operating Characteristic-Area under Curve (ROC-AUC) and Precision-Recall Curve-Area under Curve (PRC-AUC) respectively. To further enhance the prediction accuracy of the model, an adaptive Huber loss function is designed to optimize the training process, resulting in an improvement of up to 11.03% in ROC-AUC compared with the baseline model. Lastly, extended experiments are conducted to study the effects of parameters and convolutional kernel size on performance, which find a better configuration.

Understanding shared bike usages toward metros with fewer physical road separations

Physical road separations are normally considered cyclist-friendly, but whether they are unfriendly to the combination of bike-share (BS) and metros is seldom investigated since they might affect the cycling’s flexibility and convenience to cross streets and reach the destination. Although integrating BS and metro is thought to mitigate traffic congestion, how traffic congestion affects the usage of integration remains unknown. This study, conducted in Suzhou, China, examines new factors (road separation and traffic congestion) alongside well-studied factors influencing BS and metro integration. The survey revealed increased BS usage frequency after a new metro’s opening. Variables such as road separations, traffic congestion information, road network density, and proximity to metros are considered. They are processed in a selected area considering cyclable network and Thiessen Polygon corresponding to the selected metro stations. An ordered probit model is established to investigate significant factors. It is found that as more columns of road separations exist, the cyclists are less likely to use BS towards metros, regardless of whether they are cycling weekends or weekdays. Interestingly, after the new metro opened and a new metro hub was formed, proximity to the new metro hub is associated with lower BS transfer demand. A higher congestion level promotes more cycling toward the metro system on weekdays as well as higher parking fees. This indicates that the combination of BS and metro could attract former motor vehicle users and this finding could be instructive for urban planners and road designers.

Strategic bidding by predicting locational marginal price with aggregated supply curve

Price-makers in the generation sector can utilize offer data published by independent system operators (ISOs) for strategic bidding. Many research uses system-wide offer information for strategic bidding because individual offers are usually published with masked identifications. However, these methods are not applicable in markets with transmission congestion because the nodal information is not used. A research gap remains in using system-wide offer information and accessible nodal information for strategic bidding in markets with congestion. System-wide aggregated supply curves can be formed just with anonymous offers, and locational marginal price (LMP) is accessible nodal data containing the congestion information. This paper proposes a framework for price-makers to bid strategically by predicting LMPs with aggregated supply curves. A feature extraction method is proposed to make aggregated supply curves applicable for LMP prediction, and the maximal information coefficient is developed for feature selection. A convolutional neural network is combined with a long short-term memory network to model the impact of aggregated supply curves on LMPs. In this framework, price-makers can investigate the impact of their strategies on aggregated supply curves, predict LMPs with aggregated supply curves, and make the optimal bidding strategies. Numerical results based on the PJM-5 bus system and real data from the Midcontinent Independent System Operator validate the effectiveness of the proposed framework.

ENHANCING TRAFFIC SAFETY: A COMPREHENSIVE APPROACH THROUGH REAL-TIME DATA AND INTELLIGENT TRANSPORTATION SYSTEMS

Today, traffic accidents are still a difficult and urgent problem for many countries around the world. Traffic accidents on highways are often more serious than accidents on urban roads. Therefore, disseminating emergency information and creating immediate connections with road users is key to rescuing passengers and reducing congestion. Thus, this study applies data fusion and data mining techniques to analyze travel time and valuable information about traffic accidents based on the real-time data collected from On-Board Unit installed in vehicles. The results show that this important information is the vital database to analyze traffic conditions and safety factors, thereby developing a smart traffic information platform. This result enables traffic managers to provide real-time traffic information or forecasts of congestion and traffic accidents to road users. This helps limit congestion and serious accidents on the Highway.

Minimizing Noise in Location Privacy Protection Through Equipment Error Consideration

In recent years, systems that collect location information and publish statistics, such as those that publish congestion information, have been extensively employed. Because it is possible to infer an individual’s identity even if the information is not directly disclosed, it is essential to disclose data with privacy protection. Therefore, privacy protection methods based on differential privacy are attracting attention. Geo-indistinguishability is the most famous extension theorem of differential privacy for location information. Geo-indistinguishability can be achieved by adding noise to a target value that must be protected. However, noise addition reduces the usefulness of the data. Thus, it is desirable to add minimal noise to your privacy budget. Therefore, we focus on the fact that the values obtained using measurement devices contain errors. We introduced a novel concept of differential privacy tailored for location information, termed true-value-based geo-indistinguishability (T-Geo-I), which accounts for equipment noise. We also proposed a location information privacy protection method that considers T-Geo-I and reduces the amount of added noise. The object of privacy protection should be the “true value” not the “measured value” that includes measurement errors. According to the experimental results, in the case wherein the measurement error is the normal distribution, our method reduced the noise average and mean square error (MSE) by up to 41% and 63%, respectively, compared with conventional methods while maintaining a prespecified level of privacy in 108 samples of numerical data. In the case wherein the measurement error is the lognormal distribution, the proposed method based on T-Geo-I succeeded in reducing the noise average and MSE by up to 60% and 67%, respectively, compared with methods based on Geo-I, while maintaining a prespecified level of privacy. These findings indicate that the proposed method can improve the usefulness of data while maintaining a prespecified degree of privacy protection.

A market-based real-time algorithm for congestion alleviation incorporating EV demand response in active distribution networks

The dynamic charging behavior of electric vehicles (EVs) is causing frequent line-overloading problems and serious power security issues. Controlled and smart charging mechanisms for EVs incorporating demand response (DR) may provide significant operational flexibility to the grid operators and reduce charging costs for EV users. However, controlling EV charging may cause inconvenience to individual EV users. A market-based mechanism is proposed in this research work that allows EV users and households to participate in the network congestion alleviation DR program online while maintaining the power quality. Consumers submit their charging requirements (e.g., charging power and charging deadlines) and load curtailment tolerances to the aggregator. The aggregator is an entity assumed to have a long-term contract with the distribution system operator (DSO) and regularly receives network congestion information on behalf of retail energy users. Lyapunov optimization (LO) framework is used to reschedule the EVs, and household loads using DC optimal power flow (DCOPF) to get the dynamic congestion cost signal (CCS). The developed strategy is tested on a modified IEEE 33-bus radial active distribution network (ADN). Simulation results illustrate that the designed algorithm is promising in mitigating network congestion by ensuring significantly fewer violations of network constraints (i.e., line limits) both in terms of capacity and frequency. It also results in less energy cost as compared to other benchmark algorithms, e.g., greedy algorithm, and provides a guarantee of meeting EV charging, and flexible household loads’ (FHL) delay tolerance constraints. The average curtailment ratio of FHL and the service delay for EV-charging requests are converged to the user-defined limits, i.e., 0.25 for curtailment ratio tolerance and 10 times-lots for EV service delay. Unlike other counterparts, the developed algorithm is especially suitable for real-time applications as the per-slot average computational cost is negligible, i.e., 0.55 s.

MEDUSA: A Multi-Resolution Machine Learning Congestion Estimation Method for 2D and 3D Global Routing

Routing congestion is one of the many factors that need to be minimized during the physical design phase of large integrated circuits. In this article, we propose a novel congestion estimation method, called MEDUSA , that consists of three parts: (1) a feature extraction and “hyper-image” encoding; (2) a congestion estimation method using a fixed-resolution convolutional neural network model that takes a tile of this hyper-image as input and makes accurate congestion predictions for a small region of the circuit; and (3) a sliding-window method for repeatedly applying this convolutional neural network on a layout, thereby producing higher-resolution congestion maps for arbitrarily large circuits. The proposed congestion estimation approach works with both 2D (collapsed) and 3D global routing. Using both quantitative metrics and qualitative visual inspection, congestion maps produced with MEDUSA show better accuracy than prior estimation techniques. Global routers typically use estimation techniques during their first router iteration and then switch to using actual congestion information extracted from the intermediate router solutions. Experimental results within the same global router infrastructure show a significant impact on quality after the first routing iteration; other estimation techniques result in an average of 22% to 54% higher initial overflow counts. This initial quality improvement carries through to the final global routing solution, with other estimation techniques needing up to 5% more routing iterations and up to 3× more runtime, on average. Compared with other global routers, MEDUSA achieves comparable wire length results and lower total overflow counts (more legal global routing solutions) and is typically faster.

Achieving Fast Convergence and High Efficiency using Differential Explicit Feedback in Data Center

Since most flows are short-lived in data center networks, fast convergence becomes very important to help the short flows effectively utilize high bandwidth. Though current explicit feedback-based transport control protocols (TCPs) provide fast convergence via fine-grained congestion information from customized switches, they unavoidably incur large traffic overhead for widely existing small packets in data center applications, resulting in suboptimal network efficiency. To solve this issue, we propose a datacenter TCP based on <b>D</b>ifferential <b>E</b>xplicit <b>C</b>ongestion <b>N</b>otification, called DECN, to achieve fast convergence without any traffic overhead. Specifically, DECN feeds rate difference between the target and current rate back to the source by using multiple consecutive packets. Besides, we propose an enhanced version DECN* which obtains the optimal number of consecutive packets according to the packet loss rate. The experimental results of NS2 simulation and testbed implementation show that DECN and its enhanced version DECN* achieve comparable fast convergence as XCP without incurring any extra feedback overhead. Compared with the state-of-the-art explicit feedback-based TCPs, they reduce the flow completion time by up to 34&#x0025; in typical data center applications.

Crowd Evacuation Under Real Data: A Crowd Congestion Control Method Based on Sensors and Knowledge Graph

Crowd congestion is an important factor affecting evacuation efficiency, and reasonable regulation of crowd congestion in the evacuation process is one way to improve crowd evacuation efficiency. Since traditional crowd evacuation simulation methods are based on hypothetical scenarios and rules, crowd evacuation simulations lack realism. To solve this problem and reduce the scale of crowd congestion during evacuation, this paper proposes a crowd congestion control method based on sensors and knowledge graph. First, we model the evacuation scenario by extracting information from real scenes in a video to improve the realism of the evacuation simulation. Second, we construct a crowd congestion knowledge graph (CCKG) to represent the congestion information in evacuation scenes, which improves the model’s ability to characterize crowd information. Then, a gravity field is generated based on the information in the knowledge graph, and the crowd is guided by the gravity field for evacuation. Finally, we use the constructed CCKG to predict the next moment of congestion and regulate the crowd evacuation route in real-time to reduce crowd congestion. The experimental results show that using sensors to extract data from real scenes can improve the realism of crowd evacuation simulation. At the same time, the introduction of a knowledge graph can improve the characterization ability and prediction effect of the model, circumvent large-scale crowd congestion, and improve the efficiency of crowd evacuation.

An improved non-local awareness of congestion and load balanced algorithm for the communication of on chip 2D mesh-based network

Due to advancements in multi-core design technology, IC (Integrated Circuits) designers have expanded the single chip multi-core design. A privileged way of communication effectively between these multi-cores is a Network on-chip (NoC). Design of an effective routing algorithm capable of routing data to non-congested paths is the most notable research challenge in NoC, by retrieving congestion information of non-local nodes. This research proposed an improved congestion-aware load balancing routing algorithm. Non-local or distant links congestion awareness is done by propagating congestion information via data packets. By counting number of hops from the source node, in the quadrant of the destination node, an intermediate node has been defined, and after the calculation of the least congested route to the intermediate node, this route is also stored in the data packet for source routing. Furthermore, for load balancing network is partitioned into two areas called high congested area (HCA) and low congested area (LCA). For load balancing, from HCA a node in LCA is selected as output for data packets. Comparison of the proposed algorithm is done in the form of average latency, average throughput, power consumption, and scalability analysis under synthetic traffic patterns. Under simulation experiments, it is shown improvement in an average latency and throughput of the proposed algorithm is 31.28% and 5.28% respectively, than existing.

Time-dependent fleet size and mix multi-depot vehicle routing problem

This paper considers an extension of classical distribution problems and introduces the time-dependent fleet size and mix multi-depot vehicle routing problem. This important class of problems finds applications in urban logistics and service design. The solution to this problem impacts the performance of distribution companies and can help design policies to improve traffic and congestion issues. We propose a mathematical model for this challenging problem, along with several generic and problem-specific valid inequalities. A powerful matheuristic is proposed to solve large instances of the problem generated from real traffic data. Our matheuristic is also assessed on a set of instances from the literature. The computational results demonstrate that our two approaches can select the appropriate vehicle type to reduce the fixed costs while designing the routes of each vehicle from each selected depot, taking into account the departure time to avoid traffic congestion and minimize the total distribution costs. The results also show that our matheuristic is more effective than the exact method in terms of solution quality and computational time. Finally, we demonstrate the importance of considering congestion information in the design of the algorithm, showing that a constant travel time solution underestimates the costs by 15% and, more importantly, that an algorithm that takes traffic data into account produces solutions about 5% better than the former.

Congestion Control Using In-Network Telemetry for Lossless Datacenters

In the Ethernet lossless Data Center Networks (DCNs) deployed with Priority-based Flow Control (PFC), the head-of-line blocking problem is still difficult to prevent due to PFC triggering under burst traffic scenarios even with the existing congestion control solutions. To address the head-of-line blocking problem of PFC, we propose a new congestion control mechanism. The key point of Congestion Control Using In-Network Telemetry for Lossless Datacenters (ICC) is to use In-Network Telemetry (INT) technology to obtain comprehensive congestion information, which is then fed back to the sender to adjust the sending rate timely and accurately. It is possible to control congestion in time, converge to the target rate quickly, and maintain a near-zero queue length at the switch when using ICC. We conducted Network Simulator-3 (NS-3) simulation experiments to test the ICC’s performance. When compared to Congestion Control for Large-Scale RDMA Deployments (DCQCN), TIMELY: RTT-based Congestion Control for the Datacenter (TIMELY), and Re-architecting Congestion Management in Lossless Ethernet (PCN), ICC effectively reduces PFC pause messages and Flow Completion Time (FCT) by 47%, 56%, 34%, and 15.3×, 14.8×, and 11.2×, respectively.

State Aggregation for Distributed Value Iteration in Dynamic Programming

We propose a distributed algorithm to solve a dynamic programming problem with multiple agents, where each agent has only partial knowledge of the state transition probabilities and costs. We provide consensus proofs for the presented algorithm and derive error bounds of the obtained value function with respect to what is considered as the "true solution" obtained from conventional value iteration. To minimize communication overhead between agents, state costs are aggregated and shared between agents only when the updated costs are expected to influence the solution of other agents significantly. We demonstrate the efficacy of the proposed distributed aggregation method to a large-scale urban traffic routing problem. Individual agents compute the fastest route to a common access point and share local congestion information with other agents allowing for fully distributed routing with minimal communication between agents.

Effect of congestion avoidance due to congestion information provision on optimizing agent dynamics on an endogenous star network topology

This study elucidates the effect of congestion avoidance of agents given congestion information on optimizing traffic in a star topology in which the central node is connected to isolated secondary nodes with different preferences. Each agent at the central node stochastically selects a secondary node by referring to the declining preferences based on the congestion rate of the secondary nodes. We investigated two scenarios: (1) repeated visits and (2) a single visit for each node. For (1), we found that multivariate statistics describe well the nonlinear dependence of agent distribution on the number of secondaries, demonstrating the existence of the number of secondaries that makes the distribution the most uniform. For (2), we discovered that congestion avoidance linearizes the travel time for all agents visiting all nodes; in contrast, the travel time increases exponentially with secondaries when not referring to congestion information. Health examination epitomizes this finding; by allowing patients to be preferentially selected for examination in vacant examination sites, we can linearize the time it takes for everyone to complete their examination. We successfully described the optimization effect of congestion avoidance on the collective dynamics of agents in star topologies.

RSLB: Robust and Scalable Load Balancing in Software-Defined Data Center Networks

Data center networks demand high-performance, robust, and scalable load balancing protocols. Despite progress, existing work still cannot meet these requirements well. Software defined networking (SDN) can bring considerable flexibility to the management of data center networks. In the software-defined data center network, we design, analyze, and evaluate RSLB, a robust and scalable load balancing protocol that overcomes these challenges. RSLB uses fine-grained flowcell as the transmission unit, and uses link delay as the congestion metric. It uses a three-step routing strategy to route flowcells to the path with the least congestion. Through global congestion awareness, RSLB reduces flow completion time (FCT), and is more robust to topological asymmetries compared to existing congestion-agnostic schemes. To collect and store congestion information, RSLB adopts a distributed control structure that monitors the congestion of the entire network through multiple controllers, which makes it much more scalable for implementation in large-scale networks compare to existing congestion-aware schemes. The simulation results show that RSLB can achieve lower FCT for mice flows and higher throughput for elephant flows than existing schemes, no matter in failure-free topology or asymmetric topology.

IEACC: An Intelligent Edge-Aided Congestion Control Scheme for Named Data Networking With Deep Reinforcement Learning

As a promising implementation of Information-Centric Networking (ICN), Named Data Networking (NDN) has potential advantages over the TCP/IP network in content distribution, mobility support, etc. However, the research on NDN is still in its infancy, and congestion control, NDN’s most important functional element, poses many challenges, such as congestion detection, excessive window reduction for non-congested paths, and unfairness. In this paper, we propose an Intelligent Edge-Aided Congestion Control (IEACC) scheme for the NDN network based on Deep Reinforcement Learning (DRL). The proposed IEACC provides a proactive congestion detector that utilizes intermediate routers to transmit accurate congestion information along the path to consumers through data packets. Furthermore, considering the multi-source transmission in NDN, IEACC divides data packets into different congestion degrees by a lightweight clustering algorithm and provides suitable inputs for DRL, thereby obtaining a reasonable transmission rate. Then, it distributes the estimated bandwidth resources to consumers with transmission needs to maintain fairness. Finally, we implement our proposed scheme in the simulation platform and evaluate the performance in different scenarios. The results show that it can improve data transmission rate, reduce packet loss, and maintain fairness compared with others.

Dynamic Traveling Route Planning Method for Intelligent Transportation Using Incremental Learning-Based Hybrid Deep Learning Prediction Model with Fine-Tuning

Abstract Predicting the most favorable traveling routes for Vehicles plays an influential role in Intelligent Transportation Systems (ITS). Shortest Traveling Routes with high congestion grievously affect the driving comfort level of VANET users in populated cities. As a result, increase in journey time and traveling cost. Predicting the most favorable traveling routes with less congestion is imperative to minimize the driving inconveniences. A major downside of existing traveling route prediction models is to continuously learn the real-time road congestion data with static benchmarking datasets. However, learning the new information with already learned data is a cumbersome task. The main idea of this paper is to utilize incremental learning on the Hybrid Learning-based traffic Congestion and Timing Prediction (HL-CTP) to select realistic, congestion-free, and shortest traveling routes for the vehicles. The proposed HL-CTP model is decomposed into three steps: dataset construction, incremental and hybrid prediction model, and route selection. Firstly, the HL-CTP constructs a novel Traffic and Timing Dataset (TTD) using historical traffic congestion information. The incremental learning method updates the novel real-time data continuously with the TDD during prediction to optimize the performance efficiency of the hybrid prediction model closer to real-time. Secondly, the hybrid prediction model with various deep learning models performs better by taking the route prediction decision based on the best sub-predictor results. Finally, the HL-CTP selects the most favorable vehicle routes selected using traffic congestion, timing, and uncertain environmental information and enhances the comfort level of VANET users. In the simulation, the proposed HL-CTP demonstrates superior performance in terms of Mean Square Error (MSE), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE).

Edge-Based Congestion-Aware Datacenter Load Balancing with Smart Probing

Abstract Data center networks employ multipath topologies for delivering high bisection bandwidth. Load balancing through these multiple paths reduces latency and improves throughput. Network fabric-based schemes need customized hardware and cannot be implemented on commodity hardware. The existing end-host-based solutions have shown performance better than equal-cost multi-path (ECMP) but they use active probing and uneven congestion signal (e.g. coarse-grained RTT or ECN) that degrades performance especially in asymmetric topologies. Therefore, most of these end-host solutions cannot accurately sense congestion. End-to-end latency can effectively provide accurate path congestion information and smart probing with the help of congestion window can reduce network overhead. In this paper, we propose an edge-based congestion-aware load-balancing solution with smart probing. We combine ECN and end-to-end latency to get more accurate congestion metric and with probabilistic approach we avoid congestion at the best available path. Furthermore, we employ congestion window trend to perform smart probing that will reduce network overhead by performing probing only when required. The proposed scheme is implemented in network simulator ns-3 and flow completion time is taken as the key performance metric. The results show that the proposed scheme not only outperformed ECMP, Clove and LetFlow but also improved the performance over Hermes and CONGA at least 10–30% by employing (i) an accurate congestion signal that effectively detects congestion and asymmetries, and (ii) smart probing to reduce network overhead.

Aberrant Driving Behavior Prediction for Urban Bus Drivers in Taiwan Using Heart Rate Variability and Various Machine Learning Approaches: A Pilot Study

Objective: Aberrant driving behavior (ADB) decreases road safety and is particularly relevant for urban bus drivers, who are required to drive daily shifts of considerable duration. Although numerous frameworks based on human physiological features have been applied to predict ADB, the research remains at an early stage. This study used heart rate variability (HRV) parameters to establish ADB occurrence prediction models with various machine learning approaches. Methods: Twelve Taiwanese urban bus drivers were recruited for four consecutive days of naturalistic driving data collection (from their routine routes) between March and April 2020; driving behaviors and physiological signals were obtained from provided devices. Weather and traffic congestion information was determined from public data, while sleep quality and professional driving experience were self-reported. To develop the ADB prediction model, several machine learning models—logistic regression, random forest, naive Bayes, support vector machine, and gated recurrent unit (GRU)—were trained and 10-fold cross-validated by using the testing data. Results: Most drivers with ADB reported deficient sleep quality (≤80%), with significantly higher mean scores on the Karolinska Sleepiness Scale and driver behavior questionnaire subcategory of lapses and errors than drivers without ADB. Next, HRV indices significantly differed between the measurement of a pre-ADB event and a baseline. The accuracy of the GRU models ranged from 78.84% ± 1.49% to 89.57% ± 1.31%. Conclusion: Drivers with ADB tend to have inadequate sleep quality, which may increase their fatigue levels and impair driving performance. The established time-series models can be considered for ADB occurrence prediction among urban bus drivers.

A Smart Evacuation Guidance System for Large Buildings

In large buildings, during the situation of fire or other hazards, a smart evacuation guidance system needs to fully consider manifold aspects of hazards to guide evacuees through exit gates as fast as possible by dynamic and safe routes. In this paper, we propose a smart evacuation guidance system with a dynamic evacuation routing approach by using the LCDT (Length-Capacity-Density-Trustiness) weighted graph model and partial view (PV) information which represents the hazard intensity and the crowd congestion information of a group of sections/floors in the building. The proposed system is designed as a distributed system with multiple layers of computing by using smart indicators. Given such a system, we develop an efficient distributed approach, so-called LCDT&amp;PV, to find out effective evacuation routes dynamically. We then propose an estimating congestion strategy in order to improve the efficiency of dynamic evacuation routes. To validate the proposed system, we implement a simulator to compare the proposed evacuation routing approach with baseline approaches. Experimental results show that the proposed approach reduces up to 30% of the total evacuation time compared with others. Furthermore, through the results of initial smart indicator implementation, which can interact with the simulator, we show the viability of the proposed system.