Rajesh Krishnan

A computationally efficient two-stage method for short-term traffic prediction on urban roads

Abstract Short-term traffic prediction plays an important role in intelligent transport systems. This paper presents a novel two-stage prediction structure using the technique of Singular Spectrum Analysis (SSA) as a data smoothing stage to improve the prediction accuracy. Moreover, a novel prediction method named Grey System Model (GM) is introduced to reduce the dependency on method training and parameter optimisation. To demonstrate the effects of these improvements, this paper compares the prediction accuracies of SSA and non-SSA model structures using both a GM and a more conventional Seasonal Auto-Regressive Integrated Moving Average (SARIMA) prediction model. These methods were calibrated and evaluated using traffic flow data from a corridor in Central London under both normal and incident traffic conditions. The prediction accuracy comparisons show that the SSA method as a data smoothing step before the application of machine learning or statistical prediction methods can improve the final traffic prediction accuracy. In addition, the results indicate that the relatively novel GM method outperforms SARIMA under both normal and incident traffic conditions on urban roads.

On the estimation of space-mean-speed from inductive loop detector data

Abstract Travel time is an important indicator of network performance used in traffic operations and management. Commonly deployed inductive loop detectors (ILDs) measure time-mean-speed (TMS), whereas space-mean-speed (SMS) is required to calculate the travel time. A well-known relationship between the TMS and the SMS was derived by Wardrop. However, this relationship cannot be used in practice to estimate travel times as it requires knowledge of the variance of the SMS. The variance of the SMS is not measured by the ILDs and is normally not available in practice. A novel formulation is presented in this paper to estimate the SMS using TMS obtained from ILDs. In addition, two additional models based on the formulation are developed to improve the estimation performance by taking traffic states into account. The initial results show that the proposed formulation can used to estimate the SMS, and hence the travel time, accurately using real-world data.

Short-term prediction of traffic flow using a binary neural network

AbstractThis paper introduces a binary neural network-based prediction algorithm incorporating both spatial and temporal characteristics into the prediction process. The algorithm is used to predict short-term traffic flow by combining information from multiple traffic sensors (spatial lag) and time series prediction (temporal lag). It extends previously developed Advanced Uncertain Reasoning Architecture (AURA) k-nearest neighbour (k-NN) techniques. Our task was to produce a fast and accurate traffic flow predictor. The AURA k-NN predictor is comparable to other machine learning techniques with respect to recall accuracy but is able to train and predict rapidly. We incorporated consistency evaluations to determine whether the AURA k-NN has an ideal algorithmic configuration or an ideal data configuration or whether the settings needed to be varied for each data set. The results agree with previous research in that settings must be bespoke for each data set. This configuration process requires rapid and scalable learning to allow the predictor to be set-up for new data. The fast processing abilities of the AURA k-NN ensure this combinatorial optimisation will be computationally feasible for real-world applications. We intend to use the predictor to proactively manage traffic by predicting traffic volumes to anticipate traffic network problems.

Reliability of Bluetooth Technology for Travel Time Estimation

A unique Bluetooth-enabled device may be detected several times or not at all when it passes a sensor location. This depends mainly on the strength and speed of a transmitting device, discovery procedure, location of the device relative to the Bluetooth sensor, the Bluetooth sensors ping cycle (0.1 s), the size and shape of the sensors detection zone, and the time span for which the Bluetooth-enabled device is within the detection zone. The influences of size of Bluetooth sensor detection zones and Bluetooth discovery procedure on multiple detection events have been mentioned in previous research. However, their corresponding impacts on accuracy and reliability of estimated travel time have not been evaluated. In this study, a controlled field experiment is conducted to collect both Bluetooth and global positioning system (GPS) data for 1000 trips to be used as the basis for evaluation. Data obtained by GPS logger are used to calculate actual travel time, referred to as ground truth, and to geo-code the Bluetooth detection events. In this setting, reliability is defined as the percentage of devices captured per trip during the experiment. It is found that, on average, Bluetooth-enabled devices will be detected 80% of the time while passing a sensor location. The impact of location ambiguity caused by the size of the detection zone is evaluated using geo-coded Bluetooth data. Results show that more than 80% of the detection events are recorded within the range of 100 m from the sensor center line. It is also shown that short-range antennas detect Bluetooth-enabled devices in a location closer to the sensor, thus providing a more accurate travel time estimate. However, the smaller the size of the detection zone, the lower is the penetration rate, which could itself influence the accuracy of estimates. Therefore, there has to be a trade-off between acceptable level of location ambiguity and penetration rate for configuration and coverage of the antennas.

Accuracy of Travel Time Estimation Using Bluetooth Technology: Case Study Limfjord Tunnel Aalborg

AbstarctBluetooth Technology (BT) has been used as a relatively new cost-effective measurement tool for travel time. However, due to low sampling rate of BT compared to other sensor technologies, the presence of outliers may significantly affect the accuracy and reliability of travel time estimates obtained using BT. In this study, the concept of outliers and their impact on travel time accuracy are discussed. Four different estimators, namely Min-BT, Max-BT, Med-BT and Avg-BT, were used to estimate travel times using BT. By means of various estimation methods, it is tried to evaluate the impact of estimation method on the accuracy of estimated travel time using BT. Two sources of Floating Car Data (FCD) were used as the ground truth in order to quantify and evaluate the accuracy of travel time profiles obtained by BT. Three aggregation techniques including arithmetic mean, geometric mean and harmonic mean were used to construct the travel time profile using BT dataset. In order to quantify the impact of sample size on accuracy of travel time estimates, a series of sensitivity analyses are conducted. Results show that Min-BT and Med-BT are more robust in the presence of outliers in the dataset and can provide more accurate travel time estimates compared to Max-BT and Avg-BT. Moreover, implementing harmonic mean and geometric mean for travel time profile construction could significantly improve the accuracy of estimates obtained by BT.

Comparison of modelling approaches for short term traffic prediction under normal and abnormal conditions

Short-term prediction of traffic flows is an integral component of proactive traffic management systems. Prediction during abnormal conditions, such as incidents, is important for such systems. In this paper, three different models with increasing information in explanatory variables are presented. Time Delay and Recurrent Neural Networks and the k-Nearest Neighbour (kNN) algorithms are chosen as the machine learning tools in these models. The models are tested during both normal and incident conditions. The results indicate that historical patterns provide less predictive information during incidents.

Mode-Specific Travel Time Estimation Using Bluetooth Technology

The problem of mode-specific travel time estimation is mostly relevant to arterials with different travel modes, including cars, buses, cyclists, and pedestrians. Traditional travel time measurement systems such as automated number plate recognition (ANPR) cameras detect only motor vehicles and provide an estimate of their travel times. Bluetooth technology has been used as an alternative to more expensive ANPR for travel time measurements in the recent past. However, Bluetooth-sensors detect discoverable electronic devices used by all travel modes. Bluetooth-based systems currently use the time stamp of device detection events by two sensors to estimate the travel time, and there is no direct way to estimate mode-specific travel times using this approach. Hence, estimating travel time using Bluetooth technology on urban arterials without classifying the modes of detected devices could provide a biased estimate. A novel method to estimate mode-specific travel times using Bluetooth technology that is capable of estimating mode-specific travel times, specifically distinguishing between the travel time of motor vehicles and bicycles, is presented in this article. The proposed method uses information about type of detected device (class of device, CoD) and radio signal strength indication (RSSI). The proposed method also uses the travel time of the detected device and its detection pattern across the road network by multiple Bluetooth sensors to estimate the travel mode of each detected device. The accuracy of the proposed method was evaluated against the ground truth obtained by manual transcription of traffic video recordings, and was compared against travel times obtained from ANPR, a commercially deployed Bluetooth-based method, and a clustering method. The results show that the proposed method provides travel time estimates using Bluetooth with almost the same level of accuracy as ANPR under mixed traffic conditions.

Use of Low-Level Sensor Data to Improve the Accuracy of Bluetooth-Based Travel Time Estimation

Bluetooth sensors have a large detection zone compared with other static vehicle reidentification systems. A larger detection zone increases the probability of detecting a Bluetooth-enabled device in a fast-moving vehicle, yet increases the probability of multiple detection events being triggered by a single device. The latter situation could lead to location ambiguity and could reduce the accuracy of travel time estimation. Therefore, the accuracy of travel time estimation by Bluetooth technology depends on how location ambiguity is handled by the estimation method. The issue of multiple detection events in the context of travel time estimation by Bluetooth technology has been considered by various researchers. However, treatment of this issue has been simplistic. Most previous studies have used the first detection event (enter–enter) as the best estimate. No systematic analysis has been conducted to explore the most accurate method of travel time estimation with multiple detection events. In this study, different aspects of the Bluetooth detection zone, including size and impact on the accuracy of travel time estimation, were discussed. Four methods were applied to estimate travel time: enter–enter, leave–leave, peak–peak, and combined. These methods were developed on the basis of various technical considerations related to multiple detection events. A controlled field experiment was conducted to evaluate the accuracy of the methods through comparison with the ground truth travel time data measured by Global Positioning System technology. The results showed that the accuracy of the combined and peak–peak methods was higher than that of the other methods and that the employment of the first detection event did not necessarily yield the best travel time estimation.

The influence of alternative data smoothing prediction techniques on the performance of a two-stage short-term urban travel time prediction framework

ABSTRACT This article investigates the impact of alternative data smoothing and traffic prediction methods on the accuracy of the performance of a two-stage short-term urban travel time prediction framework. Using this framework, we test the influence of the combination of two different data smoothing and four different prediction methods using travel time data from two substantially different urban traffic environments and under both normal and abnormal conditions. This constitutes the most comprehensive empirical evaluation of the joint influence of smoothing and predictor choice to date. The results indicate that the use of data smoothing improves prediction accuracy regardless of the prediction method used and that this is true in different traffic environments and during both normal and abnormal (incident) conditions. Moreover, the use of data smoothing in general has a much greater influence on prediction performance than the choice of specific prediction method, and this is independent of the specific smoothing method used. In normal traffic conditions, the different prediction methods produce broadly similar results but under abnormal conditions, lazy learning methods emerge as superior.

A comparative study of k-NN and hazard-based models for incident duration prediction

The motivation behind this paper is to enhance the reliability of in-vehicle navigation systems by predicting the duration of incidents that cause congestion. The main objective of this paper is to develop a methodology for predicting incident duration using broadcast incident data and evaluate the performance of k-NN and hazard-based duration models for predicting incident duration; both of the models are presented in this paper. An incident dataset from the BBC for the Greater London area is used to evaluate the accuracy of both models so that the results give a direct comparison between the models. The strengths and weaknesses of the models are discussed in the paper based on this analysis. Results show that both k-NN and hazard based models have the potential to provide accurate incident duration prediction. While k-NN based models provided marginally more accurate prediction than hazard-based models, the hazard-based duration models can provide additional information such as delay probabilities that can be used by advanced routing and navigation algorithms. Results also show that traffic information incident feeds, such as the tpegML feed from the BBC or TMC information, can be used as a potential data source for incident duration prediction in vehicle navigation systems.