Large-scale Agent-based Simulation Research Articles

A framework for modeling human behavior in large-scale agent-based epidemic simulations

Agent-based modeling is increasingly being used in computational epidemiology to characterize important behavioral dimensions, such as the heterogeneity of the individual responses to interventions, when studying the spread of a disease. Existing agent-based simulation frameworks and platforms currently fall in one of two categories: those that can simulate millions of individuals with simple behaviors (e.g., based on simple state machines), and those that consider more complex and social behaviors (e.g., agents that act according to their own agenda and preferences, and deliberate about norm compliance) but, due to the computational complexity of reasoning involved, have limited scalability. In this paper, we present a novel framework that enables large-scale distributed epidemic simulations with complex behaving social agents whose decisions are based on a variety of concepts and internal attitudes such as sense, knowledge, preferences, norms, and plans. The proposed framework supports simulations with millions of such agents that can individually deliberate about their own knowledge, goals, and preferences, and can adapt their behavior based on other agents’ behaviors and on their attitude toward complying with norms. We showcase the applicability and scalability of the proposed framework by developing a model of the spread of COVID-19 in the US state of Virginia. Results illustrate that the framework can be effectively employed to simulate disease spreading with millions of complex behaving agents and investigate behavioral interventions over a period of time of months.

Large-scale agent-based simulation model of pedestrian traffic flows

Mobility patterns of pedestrians at a very high spatial and temporal resolution support urban planning strategies and facilitate a better understanding of the transport system. In this study, we develop an agent-based model that simulates disaggregated pedestrian traffic flows at a regional scale. This model is designed to overcome limitations of existing approaches in pedestrian traffic modelling by incorporating several decision processes that are defined by probabilistic rules. These rules include activity type, mode, and route choices. The results of a case study in Salzburg (Austria) show traffic flows concentrated in the city centre and along the river. Performed complexity analysis justifies the structure and logic of the modelled system by showing improvements in results during the stepwise implementation of model concepts. We also performed uncertainty analysis to provide information on the accuracy of the modelled results, showing strong and moderate correlations against observational data.

Equity Analysis of Freight Transportation Using a Large-Scale Agent-Based Modeling Framework

Freight transportation is one of the largest contributors to the transportation sector's greenhouse gases (GHG) and noise pollution. The adverse effects of the industry are not distributed equally due to large spatial disparities, with minority and low-income populations more prone to live near freight-related facilities and truck routes. However, equity analysis focusing on freight transport is limited since empirical analysis requires costly data collection processes, and analysis in large-scale areas is challenging. The principal objective of this study is to conduct an equity analysis of the impact of freight-related trips in the Atlanta-Chattanooga-Knoxville region using a large-scale agent-based simulation framework. The main results show that low-income (less than $52k) and more heavily non-white population neighbourhoods are disadvantaged neighbourhoods in terms of noise and GHG emissions. The results and methods presented in this paper can be used to provide tools to develop transportation equity and social justice analysis in large networks and can be used to develop policies and regulations to reduce the impact of freight transportation on disadvantaged populations.

Analysis and Visualization of High-Dimensional Dynamical Systems’ Phase Space Using a Network-Based Approach

The concept of attractors is considered critical in the study of dynamical systems as they represent the set of states that a system gravitates toward. However, it is generally difficult to analyze attractors in complex systems due to multiple reasons including chaos, high-dimensionality, and stochasticity. This paper explores a novel approach to analyzing attractors in complex systems by utilizing networks to represent phase spaces. We accomplish this by discretizing phase space and defining node associations with attractors by finding sink strongly connected components (SSCCs) within these networks. Moreover, the network representation of phase space facilitates the use of well-established techniques of network analysis to study the phase space of a complex system. We show the latter by introducing a new node-based metric called attractivity which can be used in conjunction with the SSCC as they are highly correlated. We demonstrate the proposed method by applying it to several chaotic dynamical systems and a large-scale agent-based social simulation model.

Large-scale agent-based simulations of online social networks

Large-scale agent-based simulations of online social networks

Exploring multi-modal evacuation strategies for a landlocked population using large-scale agent-based simulations

At a time when the impacts of climate change and increasing urbanization are making risk management more complex, there is an urgent need for tools to better support risk managers. One approach increasingly used in crisis management is preventive mass evacuation. However, to implement and evaluate the effectiveness of such strategy can be complex, especially in large urban areas. Modeling approaches, and in particular agent-based models, are used to support implementation and to explore a large range of evacuation strategies, which is impossible through drills. One major limitation with simulation of traffic based on individual mobility models is their capacity to reproduce a context of mixed traffic. In this paper, we propose an agent-based model with the capacity to overcome this limitation. We simulated and compared different spatio-temporal evacuation strategies in the flood-prone landlocked area of the Phúc Xá district in Hanoi. We demonstrate that the interaction between distribution of transport modalities and evacuation strategies greatly impact evacuation outcomes. More precisely, we identified staged strategies based on the proximity to exit points that make it possible to reduce time spent on road and overall evacuation time. In addition, we simulated improved evacuation outcomes through selected modification of the road network.

EVALUATION OF THE IMPACTS OF AUTONOMOUS VEHICLES ON THE MOBILITY OF USER GROUPS BY USING AGENT-BASED SIMULATION

An agent-based transport simulation model is used to examine the impacts of Autonomous Vehicles (AVs) on the mobility of certain groups of people. In the state of the art, it has been found that the researchers primarily have simulation studies focusing on the impacts of AVs on people regardless of certain groups. However, this study focuses on assessing the impacts of AVs on different groups of users, where each group is affected variously by the introduction of different penetration levels of AVs into the market. The Multi-Agent Transport Simulation (MATSim) software, which applies the co-evolutionary algorithm and provides a framework to carry out large-scale agent-based transport simulations, is used as a tool for conducting the simulations. In addition to the simulation of all travellers, 3 groups of users are selected as potential users of AVs, as follow: (1) long commuters with high-income, (2) elderly people who are retired, and (3) part-time workers. Budapest (Hungary) is examined in a case study, where the daily activity plans of the households are provided. Initially, the existing daily activity plans (i.e., the existing condition) of each group are simulated and assessed before the introduction of AVs into the market. After that, the AVs are inserted into the road network, where different fleet sizes of AVs are applied based on the demand of each group. The marginal utility of the travel time spent in case of a transport mode, the AV fleet size, and the cost of the travel are the key variables that determine the use of a transport mode. The key variables are set based on the characteristics of the case study (i.e., demand) and the AVs. The results of the simulations suggest that the AVs have different degrees of influences on certain groups as demonstrated in the occurred changes on the modal share. The value of changes depends on the Value of Travel Time (VOT) of people and the used fleet size of AVs. Moreover, the influence of the traveller’s characteristics on the AVs is manifested, such as different values of fleet utilization. Furthermore, the study demonstrates that an increase in the fleet size of AVs beyond 10% of the demand does not significantly raise the modal share of AVs. The outcome of this paper might be used by decision-makers to define the shape of the AVs’ use and those groups who are interested in using AVs.

Effects of population sampling on agent-based transport simulation of on-demand services

Large-scale agent-based simulations require higher computing resources than are usually available. Consequently, many applications rely on downscaling, that is, simulating with smaller population samples in which the results are then scaled. Existing studies have shown a need to investigate the impact of downscaling on the output statistics of such simulations. Downscaling is a common practice in transport modeling. In this study, we investigate the impacts of population downscaling on a ride-sharing service with a focus on vehicle occupancy and wait time, travel time and detour time. Our findings reveal that if transport modelers want to model on-demand services with ride sharing, it is strongly recommended to use a 100% population, or when using a smaller population sample, to estimate the relative biases of their desired metrics compared to the results of a 100% population in order for their results to be applicable for real-world situations.

Modeling non-pharmaceutical interventions in the COVID-19 pandemic with survey-based simulations.

Governments around the globe use non-pharmaceutical interventions (NPIs) to curb the spread of coronavirus disease 2019 (COVID-19) cases. Making decisions under uncertainty, they all face the same temporal paradox: estimating the impact of NPIs before they have been implemented. Due to the limited variance of empirical cases, researchers could so far not disentangle effects of individual NPIs or their impact on different demographic groups. In this paper, we utilize large-scale agent-based simulations in combination with Susceptible-Exposed-Infectious-Recovered (SEIR) models to investigate the spread of COVID-19 for some of the most affected federal states in Germany. In contrast to other studies, we sample agents from a representative survey. Including more realistic demographic attributes that influence agents' behavior yields accurate predictions of COVID-19 transmissions and allows us to investigate counterfactual what-if scenarios. Results show that quarantining infected people and exploiting industry-specific home office capacities are the most effective NPIs. Disentangling education-related NPIs reveals that each considered institution (kindergarten, school, university) has rather small effects on its own, yet, that combined openings would result in large increases in COVID-19 cases. Representative survey-characteristics of agents also allow us to estimate NPIs' effects on different age groups. For instance, re-opening schools would cause comparatively few infections among the risk-group of people older than 60 years.

Optimal fleet size for a shared demand-responsive transport system with human-driven vs automated vehicles: A total cost minimization approach

Vehicle automation is expected to reduce the cost of shared demand-responsive transport (DRT) services. In this context, questions regarding the conditions under which fixed-route public transport can be replaced with shared on-demand services have emerged. The expected increase in competitiveness between fixed-route and on-demand services requires the development of frameworks that enable the analysis of transportation costs of alternative modes. In this research, we develop a total cost minimization model for demand-responsive door-to-door shared transportation services, including operator and user costs. Optimization variables are vehicle size and fleet size for operation with human-driven and automated vehicles. A hybrid approach is used in which the relevant variables are analytically and numerically modeled, using data from a large-scale agent-based simulation applied to the city of Munich. We compare the case in which all trip requests must be served with the case in which request rejections are allowed, based on waiting and travel times. Different demand levels and alternative scenarios for vehicle automation are analyzed. The results indicate that the performance of door-to-door on-demand shared systems depends on the operational scheme selected. For the DRT setup and vehicle assignment strategy studied, we find that if the system is forced to have no trip rejections, economies of scale are not present, and the high user costs hinder the system’s competitiveness even under the assumption of automated vehicles. In contrast, in a system that allows for trip rejections, economies of scale are present, and vehicle automation can especially reduce operator costs, increasing the system’s competitiveness against other transportation modes. Therefore, in our setting, the efficiency of the demand-responsive service depends on the ability to reject customers, which is against the spirit of a truly public transportation service. On scenario analysis, we show that a theoretical improvement in the performance of the real-time vehicle assignment strategy can significantly reduce total cost, with economies of scale under no-rejection operation. Future research needs to address whether the actual application of more complex vehicle assignment strategies can indeed make DRT systems more cost competitive while serving all trip requests.

Adaptivity in multi-level traffic simulation using experimental frames

Balancing between simulation performance and simulation accuracy is difficult. Most simulators focus on optimizing only one of these two aspects. Which comes at the cost of the other. In this work we demonstrate a framework to dynamically balance between computational performance and simulation accuracy based on the context of the simulation. This allows to maximize both performance and accuracy when possible during the execution of the simulation. To do this, we present a method to apply adaptive abstraction in a large-scale agent-based traffic simulation. This method uses the concept of experimental frames for simulation models in order to keep track of model validity given the current simulation context. Furthermore, we present a custom developed state-of-the-art multi-level traffic simulator that includes adaptive abstraction in the core of its simulation architecture. We validate the proposed methods in a realistic multi-level traffic simulation use case executed in an urban environment.

Returning to a Normal Life via COVID-19 Vaccines in the United States: A Large-scale Agent-Based Simulation Study.

BackgroundIn 2020, COVID-19 has claimed more than 300,000 deaths in the United States alone. Although nonpharmaceutical interventions were implemented by federal and state governments in the United States, these efforts have failed to contain the virus. Following the Food and Drug Administration's approval of two COVID-19 vaccines, however, the hope for the return to normalcy has been renewed. This hope rests on an unprecedented nationwide vaccine campaign, which faces many logistical challenges and is also contingent on several factors whose values are currently unknown.ObjectiveWe study the effectiveness of a nationwide vaccine campaign in response to different vaccine efficacies, the willingness of the population to be vaccinated, and the daily vaccine capacity under two different federal plans. To characterize the possible outcomes most accurately, we also account for the interactions between nonpharmaceutical interventions and vaccines through 6 scenarios that capture a range of possible impacts from nonpharmaceutical interventions.MethodsWe used large-scale, cloud-based, agent-based simulations by implementing the vaccination campaign using COVASIM, an open-source agent-based model for COVID-19 that has been used in several peer-reviewed studies and accounts for individual heterogeneity and a multiplicity of contact networks. Several modifications to the parameters and simulation logic were made to better align the model with current evidence. We chose 6 nonpharmaceutical intervention scenarios and applied the vaccination intervention following both the plan proposed by Operation Warp Speed (former Trump administration) and the plan of one million vaccines per day, proposed by the Biden administration. We accounted for unknowns in vaccine efficacies and levels of population compliance by varying both parameters. For each experiment, the cumulative infection growth was fitted to a logistic growth model, and the carrying capacities and the growth rates were recorded.ResultsFor both vaccination plans and all nonpharmaceutical intervention scenarios, the presence of the vaccine intervention considerably lowers the total number of infections when life returns to normal, even when the population compliance to vaccines is as low as 20%. We noted an unintended consequence; given the vaccine availability estimates under both federal plans and the focus on vaccinating individuals by age categories, a significant reduction in nonpharmaceutical interventions results in a counterintuitive situation in which higher vaccine compliance then leads to more total infections.ConclusionsAlthough potent, vaccines alone cannot effectively end the pandemic given the current availability estimates and the adopted vaccination strategy. Nonpharmaceutical interventions need to continue and be enforced to ensure high compliance so that the rate of immunity established by vaccination outpaces that induced by infections.

COVID-19 in South Africa: outbreak despite interventions

The future dynamics of the Corona Virus Disease 2019 (COVID-19) outbreak in African countries is largely unclear. Simultaneously, required strengths of intervention measures are strongly debated because containing COVID-19 in favor of the weak health care system largely conflicts with socio-economic hardships. Here we analyze the impact of interventions on outbreak dynamics for South Africa, exhibiting the largest case numbers across sub-saharan Africa, before and after their national lockdown. Past data indicate strongly reduced but still supracritical growth after lockdown. Moreover, large-scale agent-based simulations given different future scenarios for the Nelson Mandela Bay Municipality with 1.14 million inhabitants, based on detailed activity and mobility survey data of about 10% of the population, similarly suggest that current containment may be insufficient to not overload local intensive care capacity. Yet, enduring, slightly stronger or more specific interventions, combined with sufficient compliance, may constitute a viable option for interventions for South Africa.

Integrated Approach for the Assessment of Strategies for the Decarbonization of Urban Traffic

This paper presents a new methodology to derive and analyze strategies for a fully decarbonized urban transport system which combines conceptual vehicle design, a large-scale agent-based transport simulation, operational cost analysis, and life cycle assessment for a complete urban region. The holistic approach evaluates technical feasibility, system cost, energy demand, transportation time, and sustainability-related impacts of various decarbonization strategies. In contrast to previous work, the consequences of a transformation to fully decarbonized transport system scenarios are quantified across all traffic segments, considering procurement, operation, and disposal. The methodology can be applied to arbitrary regions and transport systems. Here, the metropolitan region of Berlin is chosen as a demonstration case. The first results are shown for a complete conversion of all traffic segments from conventional propulsion technology to battery electric vehicles. The transition of private individual traffic is analyzed regarding technical feasibility, energy demand and environmental impact. Commercial goods, municipal traffic and public transport are analyzed with respect to system cost and environmental impacts. We can show a feasible transition path for all cases with substantially lower greenhouse gas emissions. Based on current technologies and today’s cost structures our simulation shows a moderate increase in total systems cost of 13–18%.

An Analysis of Uniswap markets

Uniswap---and other constant product markets---appear to work well in practice despite their simplicity. In this paper, we give a simple formal analysis of constant product markets and their generalizations, showing that, under some common conditions, these markets must closely track the reference market price. We also show that Uniswap satisfies many other desirable properties and numerically demonstrate, via a large-scale agent-based simulation, that Uniswap is stable under a wide range of market conditions.

Evaluating the systemic effects of automated mobility-on-demand services via large-scale agent-based simulation of auto-dependent prototype cities

Abstract The growing demand for urban mobility highlights the need for relevant and sustainable solutions in cities worldwide. Thus, we develop and implement a framework to analyze the systemic impacts of future urban mobility trends and policies. We build on prior work in classifying the world’s cities into 12 urban typologies that represent distinct land-use and behavioral characteristics by introducing a generalized approach for creating a detailed, simulatable prototype city that is representative of a given typology. We then generate and simulate two auto-dependent (largely US-specific) prototype cities via a state-of-the-art agent-based platform, SimMobility, for integrated demand microsimulation and supply mesoscopic simulation. We demonstrate the framework by analyzing the impacts of automated mobility on-demand (AMoD) implementation strategies in the cities based on demand, congestion, energy consumption and emissions outcomes. Our results show that the introduction of AMoD cannibalizes mass transit while increasing vehicle kilometers traveled (VKT) and congestion. In sprawling auto-dependent cities with low transit penetration, the congestion and energy consumption effects under best-case assumptions are similar regardless of whether AMoD competes with or complements mass transit. In dense auto-dependent cities with moderate transit modeshare, the integration of AMoD with transit yields better outcomes in terms of VKT and congestion. Such cities cannot afford to disinvest in mass transit, as this would result in unsustainable outcomes. Overall, this framework can provide insights into how AMoD can be sustainably harnessed not only in low-density and high-density auto-dependent cities, but also in other typologies.

Dynamics of taxi-like logistics systems: theory and microscopic simulations

ABSTRACT We study the dynamics of a class of bi-agent logistics systems consisting of two types of agents interacting on an arbitrary complex network. By approximating the system with simple microscopic models and solving them analytically, we reveal some universal dynamical features and propose applications of such features for system optimisations. Large-scale agent-based numerical simulations are also carried out to explore more realistic and complicated systems, with interesting emergent behaviours that can be well understood from our analytical studies. Using the taxi system as a typical logistics system with commuters and empty taxis as two types of agents, we illustrate two dynamical phases with distinct behaviours, separated by a phase boundary that can be identified as the optimal number of taxis for a particular taxi system. These features and the tuning of the optimal number of taxis can be applied to various situations, including taxi systems allowing real-time dynamical ride-sharing.

CORMORANT

People own and carry an increasing number of ubiquitous mobile devices, such as smartphones, tablets, and notebooks. Being small and mobile, those devices have a high propensity to become lost or stolen. Since mobile devices provide access to their owners' digital lives, strong authentication is vital to protect sensitive information and services against unauthorized access. However, at least one in three devices is unprotected, with inconvenience of traditional authentication being the paramount reason. We present the concept of CORMORANT, an approach to significantly reduce the manual burden of mobile user verification through risk-aware, multi-modal biometric, cross-device authentication. Transparent behavioral and physiological biometrics like gait, voice, face, and keystroke dynamics are used to continuously evaluate the user's identity without explicit interaction. The required level of confidence in the user's identity is dynamically adjusted based on the risk of unauthorized access derived from signals like location, time of day and nearby devices. Authentication results are shared securely with trusted devices to facilitate cross-device authentication for co-located devices. Conducting a large-scale agent-based simulation of 4 000 users based on more than 720 000 days of real-world device usage traces and 6.7 million simulated robberies and thefts sourced from police reports, we found the proposed approach is able to reduce the frequency of password entries required on smartphones by 97.82% whilst simultaneously reducing the risk of unauthorized access in the event of a crime by 97.72%, compared to conventional knowledge-based authentication.

Imbalance in Mobility-on-Demand Systems

The control of large-scale mobility-on-demand systems is an emerging topic that has been considered from a system theoretical, transportation scientific, and algorithmic point of view. Existing formulations model mobility-on-demand systems in a queuing theoretical, network flow-based, or continuous, kinematic framework. In this work, we model a mobility-on-demand system as a stochastic differential equation that represents a generalization of previous approaches. Based on the model, we define system imbalance as the difference of the stochastic processes of service request arrival and vehicle arrival. We formally derive the first moment of the system imbalance for an imbalance control strategy that consists of a feedforward control approach (reference trajectory) and an additional feedback component. A distributed feedback control policy is defined that averages the imbalance across the system and therefore aims at a uniform quality of service distribution. Finally, we verify our results in a high-fidelity and large-scale agent-based simulation of a hypothetical mobility-on-demand system.

Epidemic Spreading in Urban Areas Using Agent-Based Transportation Models

Human mobility is a key element in the understanding of epidemic spreading. Thus, correctly modeling and quantifying human mobility is critical for studying large-scale spatial transmission of infectious diseases and improving epidemic control. In this study, a large-scale agent-based transport simulation (MATSim) is linked with a generic epidemic spread model to simulate the spread of communicable diseases in an urban environment. The use of an agent-based model allows reproduction of the real-world behavior of individuals’ daily path in an urban setting and allows the capture of interactions among them, in the form of a spatial-temporal social network. This model is used to study seasonal influenza outbreaks in the metropolitan area of Zurich, Switzerland. The observations of the agent-based models are compared with results from classical SIR models. The model presented is a prototype that can be used to analyze multiple scenarios in the case of a disease spread at an urban scale, considering variations of different model parameters settings. The results of this simulation can help to improve comprehension of the disease spread dynamics and to take better steps towards the prevention and control of an epidemic.