Urban time series forecasting is crucial for smart city development and is key to sustainable urban management. Although urban time series models (UTSMs) are effective in general forecasting, they often overlook low-frequency events, such as holidays and extreme weather, leading to degraded performance in practical applications. In this paper, we first investigate how UTSMs handle these infrequent patterns from a neural perspective. Based on our findings, we propose $\textbf{P}$attern $\textbf{N}$euron guided $\textbf{Train}$ing ($\texttt{PN-Train}$), a novel training method that features (i) a $\textit{perturbation-based detector}$ to identify neurons responsible for low-frequency patterns in UTSMs, and (ii) a $\textit{fine-tuning mechanism}$ that enhances these neurons without compromising representation learning on high-frequency patterns. Empirical results demonstrate that $\texttt{PN-Train}$ considerably improves forecasting accuracy for low-frequency events while maintaining high performance for high-frequency events. The code is available at https://github.com/cwang-nus/PN-Train.

The intrinsic non-stationarity of urban spatiotemporal (ST) streams, particularly unique distribution shifts that evolve over time, poses substantial challenges for accurate urban ST forecasting. Existing works often overlook these dynamic shifts, limiting their ability to adapt to evolving trends effectively. To address this challenge, we propose DOL, a novel Distribution-aware Online Learning framework designed to handle the unique shifts in urban ST streams. DOL introduces a streaming update mechanism that leverages streaming memories to strategically adapt to gradual distribution shifts. By aligning network updates with these shifts, DOL avoids unnecessary updates, reducing computational overhead while improving prediction accuracy. DOL also incorporates an adaptive spatiotemporal network with a location-specific learner, enabling it to handle diverse urban distribution shifts across locations. Experimental results on four real-world datasets confirm DOL's superiority over state-of-the-art models. The source code is available at https://github.com/cwang-nus/DOL.

Citywide spatio-temporal (ST) forecasting is a fundamental task for many urban applications, including traffic accident prediction, taxi demand planning, and crowd flow forecasting. The goal of this task is to generate accurate predictions concurrently for all regions within a city. Prior works take great effort on modeling the ST correlations. However, they often overlook intrinsic correlations and inherent data distribution across the city, both of which are influenced by urban zoning and functionality, resulting in inferior performance on citywide ST forecasting. In this paper, we introduce CityCAN, a novel causal attention network, to collectively generate predictions for every region of a city. We first present a causal framework to identify useful correlations among regions, filtering out useless ones, via an intervention strategy. In the framework, a Global Local-Attention Encoder, which leverages attention mechanisms, is designed to jointly learn both local and global ST correlations among correlated regions. Then, we design a citywide loss to constrain the prediction distribution by incorporating the citywide distribution. Extensive experiments on three real-world applications demonstrate the effectiveness of CityCAN.

Traffic management centres worldwide are facing unprecedented challenges from traffic incidents, impacting congestion, delays, pollution, and costs. Congested roads during peak hours are prone to accidents, causing significant disruptions and delays. Non-recurrent congestion such as accidents, occur unexpectedly and are major contributors to delays while disruptions in public transportation lead to increased demand for alternative modes thereby causing severe congestion. Understanding the impact of these incidents on multi-modal traffic networks is crucial for developing effective response plans. This study addresses the limitations of using only historical data to run simulations in advance. The proposed architecture enables optimized simulations with real-time inputs and accurate parameter prediction for improved response planning. This work provides a foundational proof of concept for utilizing minimal real-time data for accurate simulations of dynamic traffic behaviours through a Singapore case study. Future research can focus on comparative analysis methods, optimization strategies, and incorporating pedestrian effects. These advancements will aid transport authorities in making informed decisions for infrastructure and route planning.

The recent expansion in sensor deployment throughout cities has produced extensive traffic data, empowering the traffic management center to closely monitor traffic conditions. These sensors generate massive spatio-temporal (ST) data, serving as the resource that can be harnessed to forecast traffic conditions within the city. Accurate forecasting of these conditions is crucial for traffic management centers, which can facilitate a wide range of urban applications, such as intelligent transportation, traffic simulation, and infrastructure planning. In this summary, we focus on leveraging ST traffic data through AI techniques for forecasting applications. We design novel methodologies to extract knowledge from the traffic data, providing support for a wide variety of real-world applications, including traffic simulation. To achieve it, we introduce the ST forecasting algorithms that operate at two distinct levels: the region-based level and the road-based level. These predictions can be then utilized as input to a traffic simulation framework, enabling the generation of proactive and advantageous planning strategies for urban transportation management. At the region-based level, a city is divided into multiple regions, i.e., even grid cells. Our goal is to generate accurate predictions concurrently for all regions within a city. We first investigate the inherent characteristics of traffic data, i.e., periodicity and ST dependency. We then devise a novel periodic residual learning network (PRNet) to better model the periodicity in traffic ST data. PRNet frames traffic condition forecasting as a periodic residual learning problem by modeling the variation between past and future conditions. Compared to directly predicting traffic conditions that are highly dynamic, learning more stationary variation is much easier, which facilitates model training. Also, the learned variation enables the network to produce residuals between future conditions and its corresponding weekly observations at each time interval and thus contributes to substantially more accurate predictions. At the road-based level, we focus on modeling traffic conditions of individual roads within a region, with the aim of accurately predicting traffic conditions for each road. Among various traffic conditions, traffic incidents are important to traffic management centers, as they can cause significant disruptions and delays. Thus, we introduce an incident forecasting framework, utilizing ST forecasting algorithms, to infer the incident in advance, allowing for more proactive and responsive traffic management strategies. Specifically, we adopt the ST forecasting algorithm to generate each road's speed condition. Considering that speed conditions do not change dramatically within short intervals, we use the predicted speed to infer if a traffic incident has taken place on a road. If there is a significant deviation in the predicted speed conditions from the previous speed behavior, we generate a severity level of the incident based on the deviations. This severity level can be then integrated into simulations, allowing the system to generate response plans in advance.

Crowd flow forecasting, which aims to predict the crowds entering or leaving certain regions, is a fundamental task in smart cities. One of the key properties of crowd flow data is periodicity: a pattern that occurs at regular time intervals, such as a weekly pattern. To capture such periodicity, existing studies either fuse the periodic hidden states into channels for networks to learn or apply extra periodic strategies to the network architecture. In this paper, we devise a novel periodic residual learning network (PRNet) for a better modeling of periodicity in crowd flow data. Unlike existing methods, PRNet frames the crowd flow forecasting as a periodic residual learning problem by modeling the variation between the inputs (the previous time period) and the outputs (the future time period). Compared to directly predicting crowd flows that are highly dynamic, learning more stationary deviation is much easier, which thus facilitates the model training. Besides, the learned variation enables the network to produce the residual between future conditions and its corresponding weekly observations at each time interval, and therefore contributes to substantially more accurate multi-step ahead predictions. Extensive experiments show that PR-Net can be easily integrated into existing models to enhance their predictive performance.

Predicting the future paths of an agent's neighbors accurately and in a timely manner is central to the autonomous applications for collision avoidance. Conventional approaches, e.g., LSTM-based models, take considerable computational costs in the prediction, especially for the long sequence prediction. To support more efficient and accurate trajectory predictions, we propose a novel CNN-based spatial-temporal graph framework GraphTCN, which models the spatial interactions as social graphs and captures the spatio-temporal interactions with a modified temporal convolutional network. In contrast to conventional models, both the spatial and temporal modeling of our model are computed within each local time window. Therefore, it can be executed in parallel for much higher efficiency, and meanwhile with accuracy comparable to best-performing approaches. Experimental results confirm that our model achieves better performance in terms of both efficiency and accuracy as compared with state-of-the-art models on various trajectory prediction benchmark datasets.

Much progress has recently been made to enhance mesoscopic traffic simulation with the focus on advanced modeling features and traffic management support capabilities. Subsequently, the computational performance needs to be improved in order to make mesoscopic traffic simulation stay effective to real-time applications. This paper presents a framework for mesoscopic traffic simulation which offloads both the demand and supply components to GPU. The simulation algorithm divides the simulation flow into different steps and designs multiple kernels to handle the steps. A high level of data parallelism is achieved by assigning the GPU threads to the appropriate components of the traffic network. Several optimization options using an innovative data structure and improved warp execution are also deployed to harness the GPU performance while preserving simulation correctness. The performance of the framework is evaluated in a real network showing the speedup of up to nearly 5 times in the demand simulation and more than 4 times in the supply simulation compared to the sequential simulation.

Despite having one of the most efficient transportation systems in the world, Singapore is still faced with congestion issues regularly, especially during peak hour periods, due to a number of reasons. We investigate some of the factors contributing to this issue and propose a simulator supplied with predictive travel times through congestion prediction, in order to evaluate and improve bus utilization through effective scheduling. We introduced a conceptual framework to integrate neural network models into simulation so as to improve real-time supply based on several possibilities of demands. This paper will delineate the steps taken to produce the simulator and discuss the evaluation of these models.

Mesoscopic Traffic Simulation is an important tool in traffic analysis and traffic management support. The balance between traffic modeling details and performance has made Mesoscopic Traffic Simulation one of the key solutions for traffic controllers and policy makers. Mesoscopic traffic simulators offer acceptable speed in simulating normal traffic. However, when traffic prediction and optimization for large scale networks come into context, the performance of mesoscopic traffic simulators is unsatisfactory in optimizing a massive number of control parameters for a much longer prediction horizon. This issue again emphasizes the need to further speed up mesoscopic traffic simulation. This paper proposes a comprehensive framework to massively speed up mesoscopic traffic simulation using GPU without compromising its correctness and realistic modeling property. It also gives an in-depth analysis into the trade-off between simulation correctness and performance speedup. By combining the power of GPU with optimal design and data structures, mesoscopic traffic simulation is able to speed up to more than 6 times compared to original CPU implementation.

Multi-Objective Evolutionary Algorithms(MOEAs) have been gaining increased popularity and usage in different fields of engineering. For real world large scale optimization problems with large variable/search space, using a large population of individuals in proportion to the size of search space is ubiquitous. Solving such problems with current state of the art algorithms like NSGA-II [1] is pervasive. The strength of NSGA-II lies in its non-dominance selection procedure and non-dominance based sorting of a population of individuals. Although, the non-dominated sort is computationally efficient for a small population (102- 103) of solutions but becomes computationally expensive and slow for a large population (104- 105) of solutions. Also, various archive based algorithms [2], [3] have been proposed in past which make use of a large population apart from the principal population. Therefore, there is a huge need for a scalable and parallel implementation of NSGA-II. With advent of consumer level Graphics processing units(GPUs) and advancement of CUDA framework we try to fill this research gap using GPGPU architecture. In this paper we propose a parallel GPU based implementation of NSGA-II with major focus on non-dominated sorting. The proposed approach can be easily coupled with the original form of NSGA-II to solve real world problems using large populations.

Performance is one of the major concerns in large-scale parallel microscopic traffic simulations. This paper focuses on one of the most time-costly data structures: the two-dimensional spatial index. A drawback of using popular two-dimensional tree-based spatial indexes (e.g. the R*-Tree) in large-scale microscopic traffic simulation is the heavy cost to rebalance the tree structure when a large number of vehicles frequently update their locations. This heavy location update cost also reduces the scalability of parallel microscopic traffic simulations. We observe that in real-world traffic systems the road density during a short period is stable, which is not sensitive to an individual vehicle's location. Thus, why not build a balanced tree structure based on the average road density in a road network? Motivated by this observation, this paper proposes Sim-Tree. The key feature of the Sim-Tree is that there is no need to check or rebalance its tree structure when individual vehicles frequently update their locations. In addition, a rebalance function and a bottom-up region query function are designed to optimize Sim-Tree's region query operations. The results of experiments simulating a city-scale traffic scenario on a 6-core machine show that the Sim-Tree is scalable and performs significantly better than the R*-tree family of spatial indexes.

Mesoscopic traffic simulation is an important branch of technology to support offline large-scale simulation-based traffic planning and online simulation-based traffic management. One of the major concerns using mesoscopic traffic simulations is the performance, which means the required time to simulate a traffic scenario. At the same time, the GPU has recently been a success, because of its massive performance compared to the CPU. Thus, a critical question is "whether the GPU can be a potential high-performance platform for mesoscopic traffic simulations"? To the best of our knowledge, there is no clear answer in the research area. In this paper, we firstly propose a comprehensive framework to run a traditional time-stepped mesoscopic traffic simulation on CPU/GPU. Then, we design a boundary processing method to guarantee the correctness of running mesoscopic supply traffic simulations on the GPU. Thirdly, the proposed mesoscopic traffic simulation framework is demonstrated to simulate 100,000 vehicles moving on a large-scale grid road network. In this case study, running a mesoscopic supply traffic simulation on a GPU (GeForce GT 650M) gives 11.2 times speedup, compared with running the same supply simulation on a CPU core (Intel E5-2620). In the end, this paper explains the theoretical limitation of running mesoscopic supply traffic simulations on the GPU. In conclusion, regardless of high system complexity, the proposed mesoscopic traffic simulation framework on CPU/GPU provides an innovative and promising solution for high-performance mesoscopic traffic simulations.

Parallel traffic simulation is a critical component in large-scale traffic simulations and real-time traffic simulations. Dividing workloads evenly to multi-cores and multi-machines is a challenge in parallel traffic simulations. Current researches focus on map decomposition algorithms. However, without effective workload estimation algorithms, map decomposition algorithms tend to output imbalanced partitions. This paper proposes an elliptical-shaped workload estimation algorithm. The main idea of the algorithm is to assign the computational cost of a vehicle to links in an ellipse, whose centers (or foci points) are the vehicle's origin node and the destination node. The algorithm is evaluated on a test-bed using a mesoscopic traffic simulator on the Lower Westchester County network. Case studies show that the new algorithm reduces 36% of the estimation errors in current length of links based workload estimation algorithms.

Offline road network partitioning is the first step to space-parallel distributed transportation simulation. Currently, METIS is the most popular offline road network partitioning solution, but it cannot naturally formalize data distribution in various ITS applications, and cannot guarantee to minimize data exchanges between partitions. This paper introduces a hyper graph-based offline road network partitioning solution, which is suitable for future distributed transportation simulations with ITS applications. In [10], we proposed to formalize offline road network partitioning as a hyper graph partitioning problem, which makes it possible to minimize data exchanges between partitions. We then solved the hyper graph partitioning problem using hMETIS, a graph partitioning algorithm borrowed from Very Large Scale Integration (VLSI) applications. In this paper, our experiments based on Singapore road network showed that the hyper graph-based road network partitioning with ITS applications reduces data exchanges between partitions. We observed two features in data distributions in some ITS applications, which led us to develop the biased first choice (BFC) coarsening schema. Experiments show that BFC further reduces data exchanges between partitions. For distributed transportation simulations, where there are large amounts of data exchanged between partitions, especially by ITS applications, our proposal is one candidate solution to reduce the simulation time and increase the scalability.

Creating simulation models via composition of predefined and reusable components is an efficient way of reducing costs and time associated with the simulation model development process. However, in order to successfully compose models one has to solve the issues of syntactic and semantic composability of components. HLA is the most widely used architecture for distributed simulations today. It provides a simulation environment and standards for specifying simulation parts and interactions between simulation parts. But it provides little support for semantic composability. The Base Object Model (BOM) standard is an attempt to ease reusability and composition of simulation models. However, BOMs do not contain sufficient information for defining concepts and terms in order to avoid ambiguity, and provide no methods for matching conceptual models (state machines). In this paper, we present our approach for enhancement of the semantic contents of BOMs and propose a three-layer model for syntactic and semantic matching of BOMs. The semantic enhancement includes ontologies for entities, event and interactions in each component. We also present an OWL-S description for each component including the statemachines. The three-layer model consists of syntactic matching, static semantic matching and dynamic semantic matching utilising a set of rules for reasoning about the compositions. We also describe our discovery and matching rules, which have been implemented in the Jess inference engine. In order to test our approach we have defined some simulation scenarios and implemented BOMs as building blocks for development of those scenarios, one of which has been presented in this paper. Our result shows that the three-layer model is promising and can improve and simplify composition of BOM-based components.

Running a large-scale HLA-based simulation in a distributed environment may require large amounts of computing resources at geographically different locations. A simulation running for a long period of time may result in imbalance of load levels at different computing hosts, hence leading to degradation of performance. Performance typically improves if the workload can be equally distributed among the different computing hosts. However, the Run Time Infrastructure (RTI) does not have any facilities to perform load distribution to alleviate this problem. This paper describes an architecture that provides load distribution services over a multi-threaded RTI to improve the performance of HLA simulations. A new metric for measuring the workload for both optimistic and conservative simulation federates is proposed. Load distribution is achieved through federate migration by introducing mobility and reactivity to each simulation federate.

The high level architecture (HLA) is widely used in defense applications as a common framework for modeling and simulation. The federation development and execution process (FEDEP) is a generalized process to build HLA federations from scratch. However, simulation model design, implementation, testing and execution defined in the FEDEP are time consuming and expensive. The model construction environment (MCE) is introduced in this paper as a Web-based system to enhance the efficiency of the federation development by reusing the existing simulation object models (SOM) to build new federation object models (FOM). To ensure the compatibility of the SOMs in a FOM, a matching algorithm is defined and tested. An extensible elements scheme is designed to help shape the characteristics of the FOM/SOMs. This study can contribute to more cost-efficient methodologies for the development and execution of simulation models.

The Data Distribution Management (DDM) service in the runtime infrastructure of high level architecture is employed to reduce the amount of irrelevant data communicated between federates and cut network communication cost. Current DDM schemes include mainly region-based and grid-based filtering mechanisms. We propose an agent-based DDM filtering mechanism to use agents to perform accurate data filtering in DDM. The agent-based DDM is supposed to filter out all the irrelevant information and route only the information that subscribers want. In this paper, an agent-based DDM mechanism is developed and incorporated into the runtime infrastructure. The efficiencies of the agent-based filtering mechanism and the other two traditional mechanisms are compared in an AWACS sensing aircraft experiment scenario. The results show that agent-based filtering mechanism communicates only the exactly relevant data and minimizes the network communication cost.