Variabla hastighetsgränser är väl utbrett på Stockholms stadsnära motorvägar och en del av Stockholms Motorway Control System (MCS). Målet med dagens system är att minska risken för olyckor och följdolyckor vid låga hastigheter, trafikstockningar m.m. Detta görs genom att mäta medelhastigheten med hjälp av fasta detektorer och uppdatera hastigheten som visas på variabla meddelandeskyltar utifrån rådande trafiktillstånd. I och med att efterfrågan på resor i Stockholm under rusningstid överstiger den tillgängliga kapaciteten i vägnätet är behovet av ett effektivt trafiksystem stort. Variabla hastighetsgränser kan bidra till ökad framkomlighet, men då dagens system har som målsättning att öka säkerheten är effekter som leder till ökad framkomlighet begränsade. Dessutom finns det i dagens system ett stort beroende av fungerande detektorer som mäter trafiktillståndet så korrekt som möjligt för att valet av hastighet ska kunna bestämmas på ett effektivt sätt.

Syftet med den här rapporten är att undersöka alternativa styralgoritmer för att bestämma variabla hastighetsgränser på Stockholms motorvägsnät. Målet är att öka framkomligheten jämfört med dagens system. Två olika sträckor med olika komplexitet, trafiksituation och problematik studeras. Valet av studerade styralgoritmer för de olika sträckorna väljs för att på bästa sätt motverka den problematiken som uppstår på de specifika sträckorna. Därmed kan val av algoritmer komma att skilja sig åt beroende på sträcka. I projektet utvärderas också om estimering av trafiktillståndet kan användas för att förbättra informationsflödet till algoritmerna då detektorer inte fungerar som de ska eller helt saknas, vilket i sin tur kan leda till förbättrad anpassning av de variabla hastighetsgränserna. Detta görs av för en av de studerade sträckorna. Styralgoritmerna utvärderas med mikroskopisk trafiksimulering och metoden utvecklad i projektet Mobile Millenium Stockholm (MMS), som bygger på en makroskopisk trafikflödesmodell och Kalman filtrering, används för estimering av trafiktillståndet.

Resultatet visar att det finns styralgoritmer med potential att öka framkomligheten. Valet av styralgoritm är dock beroende av typ av trafiksituation, vägdesignens komplexitet och trafikförhållanden på vägen. Det betyder att olika styralgoritmer kan prestera olika bra beroende på vilken vägsträcka man studerar. Vidare är estimering av trafiktillståndet användbart vid förlorad information på grund av icke-fungerande detektorer eller som komplement till detektorer för att minska mängden stationär utrustning.

Variable speed limits are commonly used on Stockholm’s urban motorways, and it is part of the Stockholm Motorway Control System (MCS). The goal of today’s system is to reduce the risk of accidents during congested conditions, traffic jams etc. This is done by updating the speed limits shown on variable message signs based on a measured average speed at fixed detectors. As the demand for travel in Stockholm during peak-hours exceeds the available capacity in the road network, the need for an efficient traffic system is high. Variable speed limit systems have the possibility to contribute to increased efficiency, but since today’s system aims to increase safety, effects that lead to increased efficiency are limited. Further, in todays’ variable speed limit systems there are a large dependency of precise and available measurements from stationary detectors to be able to display speed limits that reflects the current traffic conditions. The purpose of this report is to investigate alternative control algorithms to decide on the variable speed limits to be displayed at variable message signs on the urban motorway of Stockholm.

The goal is to increase efficiency compared to today's system. Two different road stretches with different complexity and different traffic conditions, resulting in two different types of congestion, are studied. Thereby, the studied control algorithms on the two road stretches are chosen based on the possibility of solving a specific problematic traffic situation in the best way. Hence, the studied control algorithms might differ for the two road stretches. Furthermore, for one of the roads stretches it is investigated if estimation of the traffic state can be used as input to the control algorithm as a complement to missing and erogenous measurements from stationary detectors in order to improve the calculations of the variable speed limits. The control algorithms are evaluated with microscopic traffic simulation and the method developed in the project Mobile Millenium Stockholm (MMS), using a macroscopic traffic flow model together with a Kalman filter, is used for estimation of the traffic state.

The result shows that there are control algorithms with the potential to increase efficiency. However, the choice of suitable control algorithm for improving traffic efficiency is dependent on the traffic situation, the complexity of the road design and the traffic conditions. Furthermore, estimation of the traffic state is useful when information is lost due to malfunctioning detectors or as a complement to reduce the amount of stationary detectors.

The signaling data in cellular networks provide means for analyzing the use of transportation systems. We propose methods that aim to reconstruct the used route through a transportation network from call detail records (CDRs) which are spatially and temporally sparse. The route estimation methods are compared based on the individual routes estimated. We also investigate the effect of different route estimation methods when employed in a complete network assignment for a larger city. Using an available CDR dataset for Dakar, Senegal, we show that the choice of the route estimation method can have a significant impact on resulting link flows.

Cellular networks of today generate a massive amount of signalling data. A large part of this signalling is generated to handle the mobility of subscribers and contains location information that can be used to fundamentally change our understanding of mobility patterns. However, the location data available from standard interfaces in cellular networks is very sparse and an important research question is how this data can be processed in order to efficiently use it for traffic state estimation and traffic planning.

In this thesis, the potentials and limitations of using this signalling data in the context of estimating the road network traffic state and understanding mobility patterns is analyzed. The thesis describes in detail the location data that is available from signalling messages in GSM, GPRS and UMTS networks, both when terminals are in idle mode and when engaged in a telephone call or a data session. The potential is evaluated empirically using signalling data and measurements generated by standard cellular phones. The data used for analysis of location estimation and route classification accuracy (Paper I-IV in the thesis) is collected using dedicated hardware and software for cellular network analysis as well as tailor-made Android applications. For evaluation of more advanced methods for travel time estimation, data from GPS devices located in Taxis is used in combination with data from fixed radar sensors observing point speed and flow on the road network (Paper V). To evaluate the potential in using cellular network signalling data for analysis of mobility patterns and transport planning, real data provided by a cellular network operator is used (Paper VI).

The signalling data available in all three types of networks is useful to estimate several types of traffic data that can be used for traffic state estimation as well as traffic planning. However, the resolution in time and space largely depends on which type of data that is extracted from the network, which type of network that is used and how it is processed.

The thesis proposes new methods based on integrated filtering and classification as well as data assimilation and fusion that allows measurement reports from the cellular network to be used for efficient route classification and estimation of travel times. The thesis also shows that participatory sensing based on GPS equipped smartphones is useful in estimating radio maps for fingerprint-based positioning as well as estimating mobility models for use in filtering of course trajectory data from cellular networks.

For travel time estimation, it is shown that the CEP-67 location accuracy based on the proposed methods can be improved from 111 meters to 38 meters compared to standard fingerprinting methods. For route classification, it is shown that the problem can be solved efficiently for highway environments using basic classification methods. For urban environments the link precision and recall is improved from 0.5 and 0.7 for standard fingerprinting to 0.83 and 0.92 for the proposed method based on particle filtering with integrity monitoring and Hidden Markov Models.

Furthermore, a processing pipeline for data driven network assignment is proposed for billing data to be used when inferring mobility patterns used for traffic planning in terms of OD matrices, route choice and coarse travel times. The results of the large-scale data set highlight the importance of the underlying processing pipeline for this type of analysis. However, they also show very good potential in using large data sets for identifying needs of infrastructure investment by filtering out relevant data over large time periods.

To get a better understanding of people’s mobility, cellular network signalling data including location information, is a promising large-scale data source. In order to estimate travel demand and infrastructure usage from the data, it is necessary to identify the trips users make. We present two trip extraction methods and compare their performance using a small dataset collected in Sweden. The trips extracted are compared with GPS tracks collected on the same mobiles. Despite the much lower location sampling rate in the cellular network signalling data, we are able to detect most of the trips found from GPS data. This is promising, given the relative simplicity of the algorithms. However, further investigation is necessary using a larger dataset and more types of algorithms. By applying the same methods to a second dataset for Senegal with much lower sampling rate than the Sweden dataset, we show that the choice of the trip extraction method tends to be even more important when the sampling rate is low. 

Traffic management and traffic information are essential in urban areas, and require a good knowledge about both the current and the future traffic state. Both parametric and non-parametric traffic state prediction techniques have previously been developed, with different advantages and shortcomings. While non-parametric prediction has shown good results for predicting the traffic state during recurrent traffic conditions, parametric traffic state prediction can be used during non-recurring traffic conditions such as incidents and events. Hybrid approaches, combining the two prediction paradigms have previously been proposed by using non-parametric methods for predicting boundary conditions used in a parametric method. In this paper we instead combine parametric and non-parametric traffic state prediction techniques through assimilation in an Ensemble Kalman filter. As non-parametric prediction method a neural network method is adopted, and the parametric prediction is carried out using a cell transmission model with velocity as state. The results show that our hybrid approach can improve travel time prediction of journeys planned to commence 15 to 30 minutes into the future, using a prediction horizon of up to 50 minutes ahead in time to allow the journey to be completed.

Traffic management and traffic information are essential in urban areas and require reliable knowledge about the current and future traffic state. Parametric and nonparametric traffic state prediction techniques have previously been developed with different advantages and shortcomings. While nonparametric prediction has shown good results for predicting the traffic state during recurrent traffic conditions, parametric traffic state prediction can be used during nonrecurring traffic conditions, such as incidents and events. Hybrid approaches have previously been proposed; these approaches combine the two prediction paradigms by using nonparametric methods for predicting boundary conditions used in a parametric method. In this paper, parametric and nonparametric traffic state prediction techniques are instead combined through assimilation in an ensemble Kalman filter. For nonparametric prediction, a neural network method is adopted; the parametric prediction is carried out with a cell transmission model with velocity as state. The results show that the hybrid approach can improve travel time prediction of journeys planned to commence 15 to 30 min into the future, with a prediction horizon of up to 50 min ahead in time to allow the journey to be completed

The use of mobile phone data for planning of transport infrastructure has been shown to have great potential in providing a means of analyzing the efficiency of a transportation system and assisting in the formulation of transport models to predict its future use. In this paper we describe how this type of data can be processed and used in order to act as both enablers for traditional transportation analysis models, and provide new ways of estimating travel behavior. Specifically, we propose a technique for describing the travel demand by constructing time sliced origin destination matrices which respect the level of detail available in Call Detail Records (CDR) from mobile phone use.

When analyzing large quantities of human mobility traces, the aspects of sensitivity of traces to be analyzed, and the scale at which such analysis can be accounted for is of high importance. The sensitivity implies that identifiable information must not be inferred from the data or any analysis of it. Thus, prompting the importance of maintaining privacy during or post-analysis stages. We aggregate the raw data with the goal to retain relevant information while at the same time discard sensitive user specifics, through site sequence clustering and frequent sequence extraction. These techniques have at least three benefits: data reduction, information mining, and anonymization. Further, the paper reviews the aggregation techniques with regard to privacy in a post-processing step.

The approaches presented in the paper for estimation of travel demand and route choices, and the additional privacy analysis, build a comprehensive framework usable in the processing of mobile phone data for transportation planning.

The project presented in this paper a part of the D4D-Senegal challenge.

The number and heterogeneity of traffic sensors are steadily increasing. A large part of the emerging sensors are measuring point speeds or travel times and in order to make efficient use of this data, it is important to develop methods and frameworks for fusion of point speed and travel time measurements in real-time. The proposed method combines a macroscopic traffic model and a non-linear filter with a new measurement model for fusion of travel time observations in a system that uses the velocity of cells in the network as state vector. The method aims to improve the fusion efficiency, especially when travel time observations are relatively long compared to the spatial resolution of the estimation framework. The method is implemented using the Cell Transmission Model for velocity (CTM-v) and the Ensemble Kalman Filter (EnKF) and evaluated with promising results in a test site in Stockholm, Sweden, using point speed observations from radar and travel time observations from taxis.

The velocity based cell transmission model (CTM-v) is a discrete time dynamical model that mimics the evolution of the traffic velocity field on highways. In this paper the CTM-v model is used together with an ensemble Kalman filter (EnKF) for the purpose of velocity sensor data assimilation. We present a calibration framework for the CTM-v and EnKF. The framework consists of two separate phases. The first phase is the calibration of the parameters of the fundamental diagram and the second phase is the calibration of demand and filter parameters. Results from the calibrated model are presented for a highway stretch north of Stockholm, Sweden.

Cellular positioning has been a very active research area for the last decade. Large improvement in accuracy has been made to support, for example, e-call and other location-based services. Traditionally, cellular positioning has been limited to cellular operators equipped with expensive synchronization hardware in order to achieve good accuracy. Lately, third parties have employed fingerprinting methods to enable positioning systems independent from the cellular operators. With improved available processing power, denser cellular networks, cheaper data collection and efficient pattern matching algorithms, the fingerprinting positioning methods have also gained popularity. In this paper, we analyzed the potential of using System Frame Number (SFN) - SFN observed time differences, which are traditionally used in Time Difference of Arrival (TDOA) positioning, for fingerprinting-based positioning. A field test was performed using measurements from TEMS (Telecommunication Management System) Investigation. By combining SFN-SFN observed time differences with Received Signal Code Power (RSCP) measurements, we demonstrated an improved accuracy of the fingerprinting method by 20% compared to only using RSCP measurements. The results are promising and show good potential in using SFN-SFN observed time differences for positioning based on fingerprinting.