arXiv subjects

Saif Eddin Jabari

Publications and source records attributed to Saif Eddin Jabari.

At least 37 records · Page 2Linked to original sources

arXiv2022

Stationary states in two lane traffic: insights from kinetic theory

Kinetics of dilute heterogeneous traffic on a two lane road is formulated in the framework of the Ben-Naim Krapivsky model and stationary state properties are analytically derived in the asymptotic limit. The heterogeneity is introduced as a quenched disorder in desired speeds of vehicles. The model assumes that each vehicle/platoon in a lane moves ballistically until it approaches a slow moving vehicle/platoon and then joins it. Vehicles in a platoon are assumed to escape the platoon at a constant rate by changing lanes. Each lane is assumed to have a different escape rate. As the stationary state is approached, the platoon density in the two lanes become equal, whereas the vehicle densities and fluxes are higher in the lane with lower escape rate. A majority of the vehicles enjoy a free-flow if the harmonic mean of the escape rates of the lanes is comparable to average initial flux on the road. The average platoon size is close to unity in the free-flow regime. If the harmonic mean is lower than the average initial flux, then vehicles with desired speeds lower than a characteristic speed $v^*$ still enjoy free-flow while those vehicles with desired speeds that are greater than $v^*$ experience congestion and form platoons behind the slower vehicles. The characteristic speed depends on the mean of escape times $(R=(R_1+R_{-1})/2)$ of the two lanes (represented by 1 and -1) as $v^* \sim R^{-\frac{1}{μ+2}}$, where $μ$ is the exponent of the quenched disorder distribution for desired speed in the small speed limit. The average platoon size in a lane, when $v^* \ll 1$, is proportional to $R^{\frac{μ+1}{μ+2}}$ plus a lane dependent correction. Equations for the kinetics of platoon size distribution for two-lane traffic are also studied. It is shown that a stationary state with platoons as large as road length can occur only if the mean escape rate is independent of platoon size.

physics.soc-ph↗

arXiv2022

Generalized Adaptive Smoothing Using Matrix Completion for Traffic State Estimation

The Adaptive Smoothing Method (ASM) is a data-driven approach for traffic state estimation. It interpolates unobserved traffic quantities by smoothing measurements along spatio-temporal directions defined by characteristic traffic wave speeds. The standard ASM consists of a superposition of two a priori estimates weighted by a heuristic weight factor. In this paper, we propose a systematic procedure to calculate the optimal weight factors. We formulate the a priori weights calculation as a constrained matrix completion problem, and efficiently solve it using the Alternating Direction Method of Multipliers (ADMM) algorithm. Our framework allows one to further improve the conventional ASM, which is limited by utilizing only one pair of congested and free flow wave speeds, by considering multiple wave speeds. Our proposed algorithm does not require any field-dependent traffic parameters, thus bypassing frequent field calibrations as required by the conventional ASM. Experiments using NGSIM data show that the proposed ADMM-based estimation incurs lower error than the ASM estimation.

math.OC↗

arXiv2022

PiDAn: A Coherence Optimization Approach for Backdoor Attack Detection and Mitigation in Deep Neural Networks

Backdoor attacks impose a new threat in Deep Neural Networks (DNNs), where a backdoor is inserted into the neural network by poisoning the training dataset, misclassifying inputs that contain the adversary trigger. The major challenge for defending against these attacks is that only the attacker knows the secret trigger and the target class. The problem is further exacerbated by the recent introduction of "Hidden Triggers", where the triggers are carefully fused into the input, bypassing detection by human inspection and causing backdoor identification through anomaly detection to fail. To defend against such imperceptible attacks, in this work we systematically analyze how representations, i.e., the set of neuron activations for a given DNN when using the training data as inputs, are affected by backdoor attacks. We propose PiDAn, an algorithm based on coherence optimization purifying the poisoned data. Our analysis shows that representations of poisoned data and authentic data in the target class are still embedded in different linear subspaces, which implies that they show different coherence with some latent spaces. Based on this observation, the proposed PiDAn algorithm learns a sample-wise weight vector to maximize the projected coherence of weighted samples, where we demonstrate that the learned weight vector has a natural "grouping effect" and is distinguishable between authentic data and poisoned data. This enables the systematic detection and mitigation of backdoor attacks. Based on our theoretical analysis and experimental results, we demonstrate the effectiveness of PiDAn in defending against backdoor attacks that use different settings of poisoned samples on GTSRB and ILSVRC2012 datasets. Our PiDAn algorithm can detect more than 90% infected classes and identify 95% poisoned samples.

cs.LG↗

arXiv2022

Incorporating Kinematic Wave Theory into a Deep Learning Method for High-Resolution Traffic Speed Estimation

We propose a kinematic wave-based Deep Convolutional Neural Network (Deep CNN) to estimate high-resolution traffic speed fields from sparse probe vehicle trajectories. We introduce two key approaches that allow us to incorporate kinematic wave theory principles to improve the robustness of existing learning-based estimation methods. First, we propose an anisotropic traffic kernel for the Deep CNN. The anisotropic kernel explicitly accounts for space-time correlations in macroscopic traffic and effectively reduces the number of trainable parameters in the Deep CNN model. Second, we propose to use simulated data for training the Deep CNN. Using a targeted simulated data for training provides an implicit way to impose desirable traffic physical features on the learning model. In the experiments, we highlight the benefits of using anisotropic kernels and evaluate the transferability of the trained model to real-world traffic using the Next Generation Simulation (NGSIM) and the German Highway Drone (HighD) datasets. The results demonstrate that anisotropic kernels significantly reduce model complexity and model over-fitting, and improve the physical correctness of the estimated speed fields. We find that model complexity scales linearly with problem size for anisotropic kernels compared to quadratic scaling for isotropic kernels. Furthermore, evaluation on real-world datasets shows acceptable performance, which establishes that simulation-based training is a viable surrogate to learning from real-world data. Finally, a comparison with standard estimation techniques shows the superior estimation accuracy of the proposed method.

cs.LG↗

arXiv2021

Stop-and-Go: Exploring Backdoor Attacks on Deep Reinforcement Learning-based Traffic Congestion Control Systems

Recent work has shown that the introduction of autonomous vehicles (AVs) in traffic could help reduce traffic jams. Deep reinforcement learning methods demonstrate good performance in complex control problems, including autonomous vehicle control, and have been used in state-of-the-art AV controllers. However, deep neural networks (DNNs) render automated driving vulnerable to machine learning-based attacks. In this work, we explore the backdooring/trojanning of DRL-based AV controllers. We develop a trigger design methodology that is based on well-established principles of traffic physics. The malicious actions include vehicle deceleration and acceleration to cause stop-and-go traffic waves to emerge (congestion attacks) or AV acceleration resulting in the AV crashing into the vehicle in front (insurance attack). We test our attack on single-lane and two-lane circuits. Our experimental results show that the backdoored model does not compromise normal operation performance, with the maximum decrease in cumulative rewards being 1%. Still, it can be maliciously activated to cause a crash or congestion when the corresponding triggers appear.

cs.CR↗

arXiv2021

Nonlinear Traffic Prediction as a Matrix Completion Problem with Ensemble Learning

This paper addresses the problem of short-term traffic prediction for signalized traffic operations management. Specifically, we focus on predicting sensor states in high-resolution (second-by-second). This contrasts with traditional traffic forecasting problems, which have focused on predicting aggregated traffic variables, typically over intervals that are no shorter than 5 minutes. Our contributions can be summarized as offering three insights: first, we show how the prediction problem can be modeled as a matrix completion problem. Second, we employ a block-coordinate descent algorithm and demonstrate that the algorithm converges in sub-linear time to a block coordinate-wise optimizer. This allows us to capitalize on the "bigness" of high-resolution data in a computationally feasible way. Third, we develop an ensemble learning (or adaptive boosting) approach to reduce the training error to within any arbitrary error threshold. The latter utilizes past days so that the boosting can be interpreted as capturing periodic patterns in the data. The performance of the proposed method is analyzed theoretically and tested empirically using both simulated data and a real-world high-resolution traffic dataset from Abu Dhabi, UAE. Our experimental results show that the proposed method outperforms other state-of-the-art algorithms.

math.OC↗

arXiv2021

A Real-Time Dispatching Strategy for Shared Automated Electric Vehicles with Performance Guarantees

Real-time vehicle dispatching operations in traditional car-sharing systems is an already computationally challenging scheduling problem. Electrification only exacerbates the computational difficulties as charge level constraints come into play. To overcome this complexity, we employ an online minimum drift plus penalty (MDPP) approach for SAEV systems that (i) does not require a priori knowledge of customer arrival rates to the different parts of the system (i.e. it is practical from a real-world deployment perspective), (ii) ensures the stability of customer waiting times, (iii) ensures that the deviation of dispatch costs from a desirable dispatch cost can be controlled, and (iv) has a computational time-complexity that allows for real-time implementation. Using an agent-based simulator developed for SAEV systems, we test the MDPP approach under two scenarios with real-world calibrated demand and charger distributions: 1) a low-demand scenario with long trips, and 2) a high-demand scenario with short trips. The comparisons with other algorithms under both scenarios show that the proposed online MDPP outperforms all other algorithms in terms of both reduced customer waiting times and vehicle dispatching costs.

eess.SY↗

arXiv2021

Learning Traffic Speed Dynamics from Visualizations

Space-time visualizations of macroscopic or microscopic traffic variables is a qualitative tool used by traffic engineers to understand and analyze different aspects of road traffic dynamics. We present a deep learning method to learn the macroscopic traffic speed dynamics from these space-time visualizations, and demonstrate its application in the framework of traffic state estimation. Compared to existing estimation approaches, our approach allows a finer estimation resolution, eliminates the dependence on the initial conditions, and is agnostic to external factors such as traffic demand, road inhomogeneities and driving behaviors. Our model respects causality in traffic dynamics, which improves the robustness of estimation. We present the high-resolution traffic speed fields estimated for several freeway sections using the data obtained from the Next Generation Simulation Program (NGSIM) and German Highway (HighD) datasets. We further demonstrate the quality and utility of the estimation by inferring vehicle trajectories from the estimated speed fields, and discuss the benefits of deep neural network models in approximating the traffic dynamics.

cs.LG↗

arXiv2021

On Equilibrium Metropolis Simulations on Self-Organized Urban Street Networks

Urban street networks of unplanned or self-organized cities typically exhibit astonishing scale-free patterns. This scale-freeness can be shown, within the maximum entropy formalism (MaxEnt), as the manifestation of a fluctuating system that preserves on average some amount of information. Monte Carlo methods that can further this perspective are cruelly missing. Here we adapt to self-organized urban street networks the Metropolis algorithm. The "coming to equilibrium" distribution is established with MaxEnt by taking scale-freeness as prior hypothesis along with symmetry-conservation arguments. The equilibrium parameter is the scaling; its concomitant extensive quantity is, assuming our lack of knowledge, an amount of information. To design an ergodic dynamics, we disentangle the state-of-the-art street generating paradigms based on nonoverlapping walks into layout-at-junction dynamics. Our adaptation reminisces the single-spin-flip Metropolis algorithm for Ising models. We thus expect Metropolis simulations to reveal that self-organized urban street networks, besides sustaining scale-freeness over a wide range of scalings, undergo a crossover as scaling varies -- literature argues for a small-world crossover. Simulations for Central London are consistent against the state-of-the-art outputs over a realistic range of scaling exponents. Our illustrative Watts-Strogatz phase diagram with scaling as rewiring parameter demonstrates a small-world crossover curving within the realistic window 2-3; it also shows that the state-of-the-art outputs underlie relatively large worlds. Our Metropolis adaptation to self-organized urban street networks thusly appears as a scaling variant of the Watts-Strogatz model. Such insights may ultimately allow the urban profession to anticipate self-organization or unplanned evolution of urban street networks.

physics.soc-ph↗

arXiv2021

Power laws and phase transitions in heterogenous car following with reaction times

We study the effect of reaction times on the kinetics of relaxation to stationary states and on congestion transitions in heterogeneous traffic. Heterogeneity is modeled as quenched disorders in the parameters of the car following model and in the reaction times of the drivers. We observed that at low densities, the relaxation to stationary state from a homogeneous initial state is governed by the same power laws as derived by E. Ben-Naim et al., Kinetics of clustering in traffic flow, Phys. Rev. E 50, 822 (1994). The stationary state, at low densities, is a single giant platoon of vehicles with the slowest vehicle being the leader. We observed formation of spontaneous jams inside the giant platoon which move upstream as stop-go waves and dissipate at its tail. The transition happens when the head of the giant platoon interacts with its tail, stable stop-go waves form, which circulate in the ring without dissipating. We observed that the system behaves differently when the transition density is approached from above that it does when approached from below. When the transition density is approached from below, the gap distribution behind the leader has a double peak and is fat-tailed but has a bounded support and thus the maximum gap in the system and the variance of the gap distribution tend to size-independent values. When the transition density is approached from above, the gap distribution becomes a power law and, consequently, the maximum gap in the system and the variance in the gaps diverge as a power law, thereby creating a discontinuity at the transition. Thus, we observe a phase transition of unusual kind in which both a discontinuity and a power law are observed at the transition density. These unusual features vanish in the absence of reaction time (e.g., automated driving).

physics.soc-ph↗

arXiv2020

Pay for Intersection Priority: A Free Market Mechanism for Connected Vehicles

The rapid development and deployment of vehicle technologies offer opportunities to re-think the way traffic is managed. This paper capitalizes on vehicle connectivity and proposes an economic instrument and corresponding cooperative framework for allocating priority at intersections. The framework is compatible with a variety of existing intersection control approaches. Similar to free markets, our framework allows vehicles to trade their time based on their (disclosed) value of time. We design the framework based on transferable utility games, where winners (time buyers) pay losers (time sellers) in each game. We conduct simulation experiments of both isolated intersections and an arterial setting. The results show that the proposed approach benefits the majority of users when compared to other mechanisms both ones that employ an economic instrument and ones that do not. We also show that it drives travelers to estimate their value of time correctly, and it naturally dissuades travelers from attempting to cheat.

eess.SY↗

arXiv2020

Backpressure Control with Estimated Queue Lengths for Urban Network Traffic

Backpressure (BP) control was originally used for packet routing in communications networks. Since its first application to network traffic control, it has undergone different modifications to tailor it to traffic problems with promising results. Most of these BP variants are based on an assumption of perfect knowledge of traffic conditions throughout the network at all times, specifically the queue lengths (more accurately, the traffic volumes). However, it has been well established that accurate queue length information at signalized intersections is never available except in fully connected environments. Although connected vehicle technologies are developing quickly, we are still far from a fully connected environment in the real world. This paper test the effectiveness of BP control when incomplete or imperfect knowledge about traffic conditions is available. We combine BP control with a speed/density field estimation module suitable for a partially connected environment. We refer to the proposed system as a BP with estimated queue lengths (BP-EQ). We test the robustness of BP-EQ to varying levels of connected vehicle penetration, and we compared BP-EQ with the original BP (i.e., assuming accurate knowledge of traffic conditions), a real-world adaptive signal controller, and optimized fixed timing control using microscopic traffic simulation with field calibrated data. Our results show that with a connected vehicle penetration rate as little as 10%, BP-EQ can outperform the adaptive controller and the fixed timing controller in terms of average delay, throughput, and maximum stopped queue lengths under high demand scenarios.

eess.SY↗

arXiv2020

Short-Term Traffic Forecasting Using High-Resolution Traffic Data

This paper develops a data-driven toolkit for traffic forecasting using high-resolution (a.k.a. event-based) traffic data. This is the raw data obtained from fixed sensors in urban roads. Time series of such raw data exhibit heavy fluctuations from one time step to the next (typically on the order of 0.1-1 second). Short-term forecasts (10-30 seconds into the future) of traffic conditions are critical for traffic operations applications (e.g., adaptive signal control). But traffic forecasting tools in the literature deal predominantly with 3-5 minute aggregated data, where the typical signal cycle is on the order of 2 minutes. This renders such forecasts useless at the operations level. To this end, we model the traffic forecasting problem as a matrix completion problem, where the forecasting inputs are mapped to a higher dimensional space using kernels. The formulation allows us to capture both nonlinear dependencies between forecasting inputs and outputs but also allows us to capture dependencies among the inputs. These dependencies correspond to correlations between different locations in the network. We further employ adaptive boosting to enhance the training accuracy and capture historical patterns in the data. The performance of the proposed methods is verified using high-resolution data obtained from a real-world traffic network in Abu Dhabi, UAE. Our experimental results show that the proposed method outperforms other state-of-the-art algorithms.

eess.SP↗

arXiv2020

Comparative Analysis of Economic Instruments in Intersection Operation: A User-Based Perspective

Focusing on different economic instruments implemented in intersection operations under a connected environment, this paper analyzes their advantages and disadvantages from the travelers' perspective. Travelers' concerns revolve around whether a new instrument is easy to learn and operate, whether it can save time or money, and whether it can reduce the rich-poor gap. After a comparative analysis, we found that both credit and free-market schemes can benefit users. Second-price auctions can only benefit high VOT vehicles. From the perspective of technology deployment and adoption, a credit scheme is not easy to learn and operate for travelers.

cs.CY↗

arXiv2020

A User-Based Charge and Subsidy Scheme for Single O-D Network Mobility Management

We propose a path guidance system with a user-based charge and subsidy (UBCS) scheme for single O-D network mobility management. Users who are willing to join the scheme (subscribers) can submit travel requests along with their VOTs to the system before traveling. Those who are not willing to join (outsiders) only need to submit travel requests to the system. Our system will give all users path guidance from their origins to their destinations, and collect a \emph{path payment} from the UBCS subscribers. Subscribers will be charged or subsided in a way that renders the UBCS strategy-proof, revenue-neutral, and Pareto-improving. A numerical example shows that the UBCS scheme is equitable and progressive.

eess.SY↗

arXiv2020

Traffic flow with multiple quenched disorders

We study heterogeneous traffic dynamics by introducing quenched disorders in all the parameters of Newell's car-following model. Specifically, we consider randomness in the free-flow speed, the jam density, and the backward wave speed. The quenched disorders are modeled using beta distributions. It is observed that, at low densities, the average platoon size and the average speed of vehicles evolve as power-laws in time as derived by Ben-Naim, Krapivsky, and Redner (BKR). No power-law behavior has been observed in the time evolution of the second moment of density and density distribution function indicating no equivalence between the present system and the sticky gas. As opposed to a totally asymmetric simple exclusion process (TASEP), we found no power-law behavior in the stationary gap distribution and the transition from the platoon forming phase to the laminar phase coincides with the free-flow to congestion transition and is always of first-order, independent of the quenched disorder in the free-flow speed. Using mean-field theory, we derived the gap distribution of vehicles and showed that the phase transition is always of first-order, independent of the quenched disorder in the free-flow speed corroborating the simulation results. We also showed that the transition density is the reciprocal of the average gap of vehicles in the platoon in the thermodynamic limit.

physics.soc-ph↗

arXiv2020

Noticeability Versus Impact in Traffic Signal Tampering

This paper investigates the vulnerability of urban traffic networks to cyber-attacks on traffic lights. We model traffic signal tampering as a bi-objective optimization problem that simultaneously seeks to reduce vehicular throughput in the network over time (maximize impact) while introducing minimal changes to network signal timings (minimize noticeability). We represent the Spatio-temporal traffic dynamics as a static network flow problem on a time-expanded graph. This allows us to reduce the (non-convex) attack problem to a tractable form, which can be solved using traditional techniques used to solve linear network programming problems. We show that minor but objective adjustments in the signal timings over time can severely impact traffic conditions at the network level. We investigate network vulnerability by examining the concavity of the Pareto-optimal frontier obtained by solving the bi-objective attack problem. Numerical experiments are carried to illustrate the types of insights that can be extracted from the Pareto-optimal frontier. For instance, our experiments suggest that the vulnerability of a traffic network to signal tampering is independent of the demand levels.

eess.SY↗

arXiv2020

Traffic Data Imputation using Deep Convolutional Neural Networks

We propose a statistical learning-based traffic speed estimation method that uses sparse vehicle trajectory information. Using a convolutional encoder-decoder based architecture, we show that a well trained neural network can learn spatio-temporal traffic speed dynamics from time-space diagrams. We demonstrate this for a homogeneous road section using simulated vehicle trajectories and then validate it using real-world data from NGSIM. Our results show that with probe vehicle penetration levels as low as 5\%, the proposed estimation method can provide a sound reconstruction of macroscopic traffic speeds and reproduce realistic shockwave patterns, implying applicability in a variety of traffic conditions. We further discuss the model's reconstruction mechanisms and confirm its ability to differentiate various traffic behaviors such as congested and free-flow traffic states, transition dynamics, and shockwave propagation.

physics.soc-ph↗