Abstract—Wireless communication became a key technology in the transportation domain. An increase of efficiency and safety can be achieved by connecting all traffic participants. In the railway domain wireless communication could enable new applications. In this contribution we focus on train-to- train (T2T) communication and investigate the propagation conditions. In typical railwayenvironments different types of objects along the track highly influence the propagation channel and, in turn, the performance of wireless T2T communications. Therefore, the characterization and modeling of the propagation effects are indispensable. In comparison to cellular or vehicle-to- vehicle communication, T2T propagation is hardly investigated. In this paper we present how key parameters of the multipath components (MPCs) of the channel impulse response can be estimated. The estimation is based on measurement data. We estimate delay and Doppler of single-bounce MPCs and estimate the location of the corresponding scatterers.
Therefore, a novel road surveillance system is introduced in , proposing to reuse signals from vehicular communica- tions infrastructure for passive radar application. The detection and localizationof road users is intended to be achieved by sensing the wireless propagation characteristics between the links of existing communication networks. Thereby, it is assumed that road users and other objects induce delayed and Doppler shifted multipath components (MPC), which correspond to their location and dynamics. Exploiting the network structure and evaluating the individual links allows to infer the location and velocity of road users. A similar idea is followed in , where all devices that transmit signals are considered as possible illuminators of opportunity. That means, besides static network nodes also mobile devices can be used. Since the localization accuracy strongly depends on precise location information about the individual network nodes, an incorporation of mobile nodes is challenging and requires to account for location uncertainties.
Current railway systems lack the capability to continuously localize the trains in the track network. For safe operation therefore large distances between consecutive trains are re- quired. This is becoming more and more of a problem because the increasing amount of passenger requires a higher density of trains on the tracks. Especially in urban areas this problem cannot be solved by building new infrastructure like tracks because of high costs and limited space. Further, building tracks is time consuming and hence cannot fix the already existing bottlenecks in the near future. One way to handle the increasing amount of passengers that avoids the before men- tioned issues of an infrastructure based approach is automa- tion. For automation accurate and reliable trainlocalization is crucial and can be seen as one of the key technologies. The research in this area is focusing on approaches with global navigation satellite systems (GNSS) . GNSS will be a viable localization solution in many scenarios but there are also environments like tunnels, train stations with closed roofs and urban canyons in which GNSS signals are strongly degraded or completely blocked. GNSS signals can also be jammed easily due to the low signal power on the receiver antenna. An alternative to GNSS is the approach that was introduced with the European train control system (ETCS). In the ETCS the train position is determined with a combination of an odometer and radio beacons. The odometer is used to perform dead reckoning in respect to the last radio beacon. The radio beacons are placed in the middle of the track between the rails at known positions. When a train passes a beacon the beacon number is transmitted to the train. With the beacon number the beacon position is obtained from a database. This is used to
Figure 2 is represented the general structure of capacity research forrailway systems. Capacity research can not only carry out macro- or mesoscopic evaluation of the rail- way operation in the whole railway network, but also evaluate the operation on a cer- tain occupancy element microscopic. The level of delay and punctuality inside the whole investigated area are two major items to improve quality of operation. Multiple simulations of same timetables with perturbation are used to check the robustness of timetables regarding delay propagation. The performance of the whole investigated area is another important objective to evaluate the quality of operation and the capacity of an investigated area. For microscopic evaluation of homogeneity, the performance of single occupancy elements is evaluated through several single simulations of differ- ent timetables in order to identify the bottlenecks in the whole network. (Cao 2017). Railway capacity is determined by factors which include train heterogeneity, train op- eration patterns, infrastructure layout, the speed of trains and train scheduling. It is an important factor as it determines the ability of stations to allow a low-speed trainto stop for high-speed trains through analyzing train schedules. Furthermore, the location of stations that can allow for the overtaking of slow-speed trains is also a significant ele- ment in evaluating railway capacity when heterogeneous trains are operated on a rail- way track. Therefore, line planning and the generation of timetables are key in deter- mining railway capacity as well as train allocation, train routing, rolling stock and crew scheduling. The aforementioned steps determine the overall railway networking and its real-time management. Thus, it is important to establish optimum railway planning that considers line planning and timetable generation simultaneously.
Although rail transport is extremely safe, collisions ofrailway vehicles happen occasionally. The safety overlay system described here adopts a very successful concept from aeronautics for avoiding the collision of trains . It combines three core technologies: A direct train-to-train communication system, an accurate localization system, and a cooperative situation analysis and decision support system. Each equipped train determines its track-selective position on the line as well as other relevant parameters such as the current calculated braking distance, and broadcasts them per direct train-to-train communication to all other trains in communication range. Intelligent algorithms on board the receiving trains evaluate this information and raise an alarm in case of danger.
ization in global navigation satellite system (GNSS) denied areas is proposed. The proposed localization system is long-term stable and based solely on magnetometer and odometer measurements. The system utilizes that the magnetic field shows strong and time persistent variations along a railway track. Two trains driving on the same track will observe the same magnetic field variations but with a certain shift. This shift depends on the relative position of the trains and their speed. By measuring the train speed with an odometer it becomes possible to estimate the relative position by comparing the magnetometer and odometer measurements of two trains. In this paper we use cross-correlation to obtain the relative position estimate from a batch of measurements. A subsequent Kalman filter is used to smooth the estimate and to incorporate prior knowledge of the train dynamics. We further derive the Cramer-Rao lower bound (CRLB) for the relative position estimate to investigate the theoretically achievable localization accuracy and to approximate the variance of the relative position estimate in the update step of the Kalman filter. In an evaluation the feasibility and accuracy of the approach is shown based on measurement data collected with a train driving in a rural area. The results indicate that with the proposed method the relative position can be estimated with sub-meter accuracy.
The basis of reliable communication is the knowledge of the influence of the propagation channel on the wireless transmission. Therefore, we performed a comprehensive mea- surement campaign within the Roll2Rail project with two high speed trains and the DLR RUSK channel sounder . The large measurement bandwidth and high spatial resolution of the sounding measurements allow a detailed analysis of the propagation characteristics. The data set is divided in railway station, open field and hilly environment and path loss models and large scale fading statistics are derived. We compare the derived parameters with existing T2T channel models presented in . These previous presented path loss (PL) models are based on received signal strength indication (RSSI) measurements with ITS-G5 transceivers. The new models are based on more extensive and preciser measurement data and therefore sharpen and extend the existing models.
The individual signatures have shown different characteristics: The magnetic signature is the most versatile and best suited for along-track estimation, switch way detection and parallel track identification. However, other trains and track works can disturb and change this signature. Curvature and attitude signatures showed suitable switch way discrimination, but limited capabilities for a continues along-track estimation, with exceptions at curve changes. The vibrations may be used for along- track estimation, but showed a speed dependency of the signature. The location dependency of some signatures is not exclusive: as shown with the bank angle and curvature, a curve causes the train cabin to swing and generates an own signature at similar train speeds of the different runs. In conclusion, some of the signatures depend on the sensor mounting, train speed and possibly on the train type. As a consequence, individual signatures demand for individual maps for each setup and train type. Therefore, an automated creation and maintenance of a map is needed, such as a SLAM (simultaneous localization and mapping) method forrailway tracks .
Abstract. Train-to-train communication will be the key technology for future railway operation. An increase of safety and efficiency can be achieved by exchanging data between trains via ad hoc networks. For vehicle-to-vehicle communication the European standard is intelligent transport systems (ITS-G5). The usage of this standard for railways is hardly investigated. We investigate the performance of ITS-G5 fortrain- to-train communication at high speed conditions. ITS-G5 units were in- stalled on two high speed trains and train-to-train (T2T) measurements were performed between Naples and Rome during four nights to cover different maneuvers. We present the analysis of the measurements data and resulting path loss models for tunnel and open field environments. Keywords: T2T, electronic coupling, ITS-G5, channel model, NGT
In order to evaluate the performance of the EKF proposed in Section 3.3 , the PCRLB is calculated for different scenarios within the measurement setup and compared to simulation results of the tracking filter. Three different scenarios are considered, each for a single scatterer moving with constant velocity. The initial absolute velocity is 1.41 m/s for every scenario. However, each scenario possesses an individual starting position and movement direction. Figure 2 b provides an overview over the considered scenarios. Referring to Figure 2 a, Scenario I is characterized by high localization capabilities throughout the whole trajectory. Scenario II, in contrast, crosses an area of poor localization capabilities between Tx and Rx 3 . In Scenario III, the trajectory starts in an area of poor localization capabilities and moves towards an area of very high localization capabilities in the main beam the transmitting node. For the calculation of the PCRLB as well as the EKF simulations, similar system and signal parameters as for calculating the static positioning CRLB are used. As process noise intensity for the covariance in Equation ( 14 ), a value of σ q 2 = 0.01 m 2 /s 3 is defined [ 23 ]. The filter state is initialized randomly according to the initial state covariance P(t 0 ) around the initial state x(t 0 ) given by the respective scenario. Thereby, P(t 0 ) is defined as 4 × 4 diagonal matrix, with an initial variance of 0.1 m 2 on position and 0.01 m 2 /s 2 on velocity, in both x- and y-directions. Determining the PCRLB and the EKF performance results requires performing multiple Monte Carlo runs. In each run, the system equations of Section 3.3 are simulated with different samples of the process noise for 10 s, which equals an average walking distance of approximately 14 m. Thus, the multitude of simulation runs allows for approximating the expectation in Equation ( 33 ). In order to estimate the MSE matrices in Equation ( 25 ), the measurement noise also needs to be sampled for each run. In this study, for every scenario, 5000 Monte Carlo runs are performed with 50 realizations of measurement noise. Hence, the EKF is evaluated 2.5 × 10 5 times in each scenario. Thereby, the number of Monte Carlo runs was chosen to achieve statistically stable results for the PCRLB and to achieve results for the EKF, which fluctuate only marginally compared to the absolute RMSE values.
The analysis of the IMU sensor data presents low dynam- ics for the train motion. The acceleration sensors measure train motion acceleration as well as engine and motion vibrations. The vibrations depend on motion, speed, stand- ing, engine system, undercarriage and finally the sensor placement. The vibrations have higher dynamics than the train motions and can be filtered by a low-pass filter. The gyroscopes measure small signals compared to the sensor noise. The low dynamics oftrain turn rates allow a low- pass filtering to improve the signal to noise ratio. Out of the train motion measurements, track features such as bank, bank change, slope, slope change, relative heading, curvature and basic track elements can be inferred. The low-cost MEMS IMU mounted in the cabin achieves good results in traction acceleration and yaw turn rate measurements as well as inferred track features such as curvature and bank change. Analysis with data sets from different train trips show a good repeatability of the track feature measurements. The analysis of the sensor signals and dynamics provides parameters which can be used for the design oftrain positioning systems, train motion models or simulators. The described track features can be used for further feature based localization.
The localizationof trains at its current state is based mainly on infrastructure components like balises and magnetic axle counters. This is sufficient for the current rail management sys- tems which allow trains only to drive with the absolute braking distance between them. However, this approach prohibits real- time localization and does not utilize the available capacity of the track network. In future, automated railway systems reliable and accurate localization systems will be indispens- able. For reliable localizationof trains, it can be exploited that the movement of the trains is limited by strong constrains forced on them by the track. Taking this into account the 3D localization can be reduced to the one dimensional arc length the train has driven on a certain part of the track network. For absolute positioning with global navigation satellite systems (GNSS) then only two satellites are needed for estimating the train position and the user clock error. This approach assumes that a digital map of the track network exists, that links the 1D position to a global 3D coordinate system.
The measurements are recorded sample by sample in the temporal domain with a constant frequency of 200 Hz. We are interested in location dependent correlations of the measurements and the railway environment. Within one track, the railwaylocalization problem is one dimensional and signals or features can be addressed in the metric, spatial domain. The transformation of the magnetic signal from time to spatial domain (i.e. seconds to meters) is achieved by the change of positions, respective the train speed. For our transformations we use directly the GNSS (Global Navigation Satellite System) speed measurement of the PVT (position, velocity, time) output of a GNSS receiver. The transformation of the time signal y t with time samples x t to the spatial signal y s with metric
Abstract—Railway operators for high speed trains, commuter trains or subways request for technologies to increase the density and efficiency on their rails. One of those future technologies is virtual coupling based on train-to-train (T2T) wireless communi- cation links. For the design of a wireless communication system, the knowledge of wave propagation is crucial. To characterize the propagation effects between two moving trains a T2T channel sounding measurement campaign was performed. To post-process the data, we use the Kalman enhanced super resolution tracking (KEST) algorithm to detect and track multipath components (MPCs). Using the single-input single-output measurement data recorded while the transmitter and receiver were in motion, the complex amplitude, the delay and the life time of individual MPCs can be estimated. First results of the post-processing step, the geometrical representation of MPCs using point scatterers and possibilities to identify scattering positions are presented in this paper.
but the shape of the Doppler cone is also influenced by the movement, meaning direction and velocity, of both in the 3D domain. The intersection of the two defined surfaces then contains the true location plus the ambiguities. For point scatterers the true reflection locations do not change, when the receiver or transmitter is moving, whereas location of the ambiguities change. Therefore we make use of the idea to calculate the PDF of the MPC positions and average them for the time the MPC is visible like described in . By averaging the PDFs, the ambiguities for point scatterer average out and therefore enhances the localization. If the true reflection location moves on a surface, then averaging might not enhance the localization, but the representation as PDF nevertheless helps to identify the most probable region of the true MPC location. Similar to , we represent the parameters τ and ν jointly as a bivariate Gaussian distribution
Abstract—Within the next decades the railway systems will change to fully autonomous high speed trains (HSTs). An increase in efficiency and safety and a reduction of costs would go hand in hand. Today’s centralized railway management system and established regulations can not cope with trains driving within the absolute braking distance as it would be necessary for electronic coupling or platooning maneuvers. Hence, to ensure safety and reliability, new applications and changes in the train control and management are necessary. Such changes demand new reliable control communication links between train-to-train (T2T) and future developments on train-to-ground (T2G). T2G will be covered by long term evolution-railway (LTE-R) which shall replace today’s global system for mobile communications-railway (GSM-R). The decentralized T2T communication is hardly in- vestigated and no technology has been selected. This publication focuses on the wide band propagationfor T2T scenarios and describes a extensive channel sounding measurement campaign with two HSTs. First results of T2T communication at high speed conditions in different environments are presented.
feasibility and quality of a railway compass due to distortions from ferromagnetic materials in the railway environment. On the other hand, if these distortions are dependent on the location, they can be beneficial for a localization application. This paper addresses the following questions: are the passive magnetic measurements useful fortrainlocalization? Therefore, this paper focuses on a suitability analysis oflocalization with passive magnetic signatures. The data set comprises magnetic data of two different sensor positions, recorded on a regional train at regular passenger service. This paper is a follow-up of  with an initial study of the bogie mounted sensor. We present additional measurements from a train cabin magnetometer and compare it with the bogie mounted sensor. The general benefit of a cabin mounted sensor is the easier installation and casing. This study contains a comprehensive analysis of several scenarios, such as parallel tracks, two-way tracks, and switches. Of special interest are differences of signatures at different tracks and the repeatability on same tracks and position. We also show and discuss the challenges fortrainlocalization with the passive magnetic measurements. Finally, elements oftrainlocalization based on magnetic signatures are presented with along-track estimation and switch way detection. The results are promising for further implementation in multisensor trainlocalization algorithms.
Abstract. In this paper we present first analyses and results of a comprehensive measurement campaign investigating the propagation channel in case of direct (base station free) communication between railway vehicles. The measurements cover urban, suburban and rural environments along a multifaceted regional railway network in the south of Bavaria. Beside different operational conditions like front, rear, and flank approaches of trains, we investigated several topo- logical scenarios on both, single and double track sections along the line. We will also discuss the observed characteristic changes in narrow band signal at- tenuation and Doppler spectra for passages through forests, hilly areas, stations and a tunnel.
Due to institutional and conceptual reasons action researchers, not only, but especially young researchers, often have great difficulties in writing ac- tion research. A major point is the lack of time and finance for social science in general, and especially for action research. Change processes on one hand side, reflection and writing on the other are characterized by different time structures. A researcher has little free space in action processes, which follow their own logics and time structure. Decisions have sometimes to be taken quickly, dialogues need time though finance and time for research are nor- mally limited; research projects have their beginnings and their ends, very of- ten limited to two years only. One has to acknowledge, that it is difficult un- der these restrictions to find free space for reflection and writing. If for finan- cial reasons one project is followed directly by the next one, there is too little, often not sufficient time in between for reflecting and writing.
There are several activities identified in the train: sleeping, eating, going to the toilet, and using mobile phone. In general, when the passengers are not sleeping, they are using mobile phone. The passenger tends to sleep during three periods of time. The eating activity happens once after the first sleeping period. The passenger’s going to the toilet twice during the journey. First, after two hours of journey and after 5 hours of journey. The study has identified the passenger’s activity during nine hours train journey. The result of this study can be beneficial to design acoustic environment scenario which considers the different activity.