Está en la página 1de 6

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 12, DECEMBER 2010, ISSN 2151‐9617 

HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/  
WWW.JOURNALOFCOMPUTING.ORG       55 

Conceptual Model for Cellular Network


Dependent Location Estimation (CNDLE) for
Trains in Indian Scenario
A. Gupta and Prof. S. Tapaswi

Abstract— Location estimation which uses mobile network as its skeletal framework has been a major subject of research and has drawn considerable
attention in the field of wireless communication. The major problems with Indian Railway are increase in the frequency of accidents and the lack of
punctuality. This is primarily due to the lack of a proper estimation of the train’s position at a given time. Although, the use of Global Positioning System
(GPS) is a convenient way to estimate such location, it suffers from some well known problems. During a train journey, many people carry cell phones
which receive intermittent control signals from the base stations. Using this fact, we propose a model for locating a train using the existing framework of
cellular network; thereby obviating the requirement of any additional infrastructure. Such estimation is made possible in a scenario where the train speeds
are moderate such that control signals can be exchanged periodically between the mobile and the base station.

Index Terms— Global System for Mobile Communications (GSM), Global Positioning System (GPS), Mobile Ad-hoc Networks (MANETs), Vehicular Ad-
hoc Networks (VANETs), Macro cells, Micro cells, Pico cells.

——————————  ——————————
of a clear line of sight with the tracking satellite (2)
1. INTRODUCTION dependence on weather conditions (3) expensiveness of good

D espite being the second largest rail network, transporting and reliable GPS tracking systems (4) need for maintenance.
20 million passengers and more than 2 million tons of One of the major problems when using GPS receivers in
freight daily[1][2], the current Indian Railway framework railway environments is the multiple electromagnetic
fails to provide correct running status of the trains; thereby interference (EMI) sources, due to numerous factors such as
causing inconvenience to the passengers. The Indian Railways rolling stock, locomotives, onboard power-supply
(IR) is facing new competitive threats with the expansion of installations, catenaries, electrical devices, and signaling and
the road network and cheap air fares. There is a compelling telecommunication equipment[4]. There has been hardly any
need for IR to become more customer focused by delivering analysis on the interference effects generated by railway
higher service levels which include (1) reducing accidents (2) equipment on the performance of GPS receivers located in the
improving passenger safety, (3) increasing punctuality, and (4) locomotive’s cabin. Manufacturer’s measurements generally
offering new passenger amenities[3]. The major challenge is focus on particular locomotives, but there is no established
the absence of a centralized system to track the position of norms pertaining to the maximum EMI radiation levels
railways at a particular moment of time. A lot of estimation of permitted for rail equipment in order to assure the correct
the train’s location is done manually which introduces errors operation of GPS receivers. Moreover, the set of specifications
and also causes unnecessary delays. This in turn causes grave for railway operations are abstruse. Requirements addressing
inconvenience to passengers who require frugality with time. accuracy, integrity, continuity, or availability are variegated;
One of the best solutions known till date for location being significantly different according to the kind of railway
estimation is based on the estimation using the Global industry. However, the cellular radio network often has good
Positioning System (GPS). It is a space-based global navigation coverage in most of the populated areas. Hence, using the
satellite system that provides reliable location and time existing cellular network for location estimations can be an
information anywhere on or near the Earth where there is an alternative method to provide location estimation, particularly
unobstructed line in urban areas [5].
of sight to four or more GPS satellites. Although GPS makes Location estimation based on the mobile network has received
use of satellites, is widely used for location estimations, it may considerable attention over the past few years due to its great
not provide the expected accurate results in urban areas, such potential to enable different kinds of location-based
as Hong Kong and India. This is because satellite signals are applications in the areas of logistics, tourism, and
often reflected, deflected, or blocked by skyscrapers, which entertainment. The search for a good location estimation
causes inaccurate estimations or precludes estimates. The algorithm that is based on the cellular radio network has
GPS suffers from several disadvantages which hamper its become a popular research topic for location aware
mass use in the Indian railway which include (1) requirement computing. In literature, many mobile location methods have
been proposed. These methods make use of the time of arrival
(TOA), angle of arrival (AOA), time difference of arrival
————————————————
(TDOA), and received signal strength (RSS) of signals for the
 A. Gupta is with the Indian Institute of Information Technology and
Management, Gwalior, India 474010.
proper estimation of the location of a mobile station (MS)
 Prof. S. Tapaswi is with Indian Institute of Information Technology and [6][7]. To deal with the problems of the communication delay
Management, Gwalior, India, 474010. influence on the safety location in control system, a study has
been conducted for the delay of communication with different
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 12, DECEMBER 2010, ISSN 2151‐9617 
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/  
WWW.JOURNALOFCOMPUTING.ORG       56 
system structures of control and length of data frame [8]. An handle the roaming users. The details of such a process are
architecture and methodology for executing a train control outside the scope of this research. Periodically, the cell phone
application in an extremely safe manner is presented in [9]. A sends a message to the network indicating where it is, a
safety analysis of a train control supervisor for Automatic process called Polling.
Train Operation (ATO), using timed Petri nets as the modeling
approach is described in [10]. An excellent implementation of
GPS for determining the position of trains for train safety in
Indian scenario has been developed by Indian Institute of
Technology, Kanpur under the project Satellite Imaging for
Rail Navigation (SIMRAN) [11].
This paper deals with the location estimation of train at a
given time using the existing cellular network. Such an
implementation obviates the requirement of an additional
infrastructure like the GPS. The location of a MS is roughly
known to the network as described in section 2. An
improvement to the location obtained can be done by
applying the CNDLE Model and algorithms. Eventually, this
model can be used for predicting the running status of trains Figure 1 – The current structure of GSM
and accidence avoidance.
The rest of the paper is organized as follows. Section 2
provides an overview of the current cellular network and its Each mobile phone is provided with a 10 digit telephone
functioning. Section 3 provides some assumptions that need to number known as the Mobile Identification Number (MIN),
be made for the proper functioning of the model. Section 4 and has a unique Electronic Serial Number (ESN). Also, a
describes the CNDLE model for railways and a description of unique 5-digit number known as System Identification Code
the algorithms required to implement the model is mentioned (SID) is assigned to each carrier by the Forward Control
in Section 5. Section 6 describes a theoretical comparison of Channel (FCC).When users switch on their mobile phones; it
the proposed model with the existing model for location tries to obtain a SID on the FCC. If the Mobile phone fails to
estimation of trains in Indian. Section VII concludes the paper. get link with the control channel, it displays a “no service”
message. If the Mobile phone gets the SID, it compares the
2. CURRENT CELLULAR NETWORK SID programmed in the phone with the received SID. If both
Global System for Mobile Communications (GSM) uses a SIDs match, the phone identifies that the cell it is
series of radio transmitters known as Base Stations (BS) to communicating is the part of its home system. The
connect a cell phone to a cellular network. Each BS can also be combination of a MIN and ESN creates a special digital
termed as a cell because it covers a certain range within a "signature" that includes the user’s SID. This signature is sent
discrete area. from his cell phone to the nearest BS asking if he as a
Base Stations are all interconnected, which facilitates the subscriber of a particular network is allowed to use of the
movement if users from one cell to another - a process called network. The request is passed on along the network of BSs to
handover without losing their connection. A set of Base BSC and finally the multifaceted heart of a cellular network -
Stations is connected to a particular Base Station Controller the Mobile Switching Center (MSC). The HLR contained in the
(BSC) and a group of BSCs are in turn connected to the Mobile MSC, immediately checks the special "signature" contained in
Switching Centre (MSC). The MSC is a critical part of the the request against its special subscriber database. If the
cellular network as it routes all your incoming and outgoing subscription is current, the MSC sends a message back to the
calls to and from the fixed-line networks or other cellular phone via the network of BSs that indicates that the user is
networks. The MSC also contains a critical database known as allowed to access the network. The name or code of that
the Home Location Register (HLR) which contains and network appears on the LCD screen of the cell phone.
provides the information required to authenticate, register and
locate the user as a that network's subscriber. Concurrently, 3. ASSUMPTIONS
the information in the HLR also points to which BSC the user’s This section states the various assumptions that need to be
cell phone is currently connected to, so that when the made to adequately represent the CNDLE model. We assume
network’s MSC needs to route an incoming call to his cell that for being a part the network, the trains are sufficiently
phone number, it will first check the HLR to see where he is. equipped with wireless transreceivers which can transmit in a
Moreover, each base station in the network contains a Visitor short range (100 m-200 m) for transmitting data packets
Location Register (VLR) which stores information about all the containing details of the journey such as train number,
mobiles that are currently under the jurisdiction of the MSC distance travelled timestamp. It is assumed that the
that it serves. Of all the information it stores about each MS, surroundings of the train contain at least one service enabled
the most important is the current Location Area Identity (LAI). mobile phone capable of receiving signals from the train. Such
LAI identifies under which BSC the MS is currently present. a capability can be easily integrated in the software of the
The data stored in the VLR has either been received from the mobile phone. Also, a central database is maintained which
HLR, or collected from the MS. The VLR is used primarily to contains the average time taken by different trains to travel
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 12, DECEMBER 2010, ISSN 2151‐9617 
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/  
WWW.JOURNALOFCOMPUTING.ORG       57 
various distances once the journey has commenced. Such a nt: Current number of functioning trains in a nation
database is deployed at a server dedicated specifically for tc: Current time
maintaining train status information. We shall refer to this C : Number of bits of control information required by the cellular
server as the central railway status server (CRSS). It is network.
assumed that trains travel at a moderate speed not more than N : Number of bits used that are required to represent each train
140 kmph and each train follows a similar path in each trip. number uniquely
D: Number of bits required to represent the maximum distance
possible between two stations uniquely.
4. THE CNDLE MODEL T: Number of bits required to represent the timestamp
representing tc
Location estimation in mobile ad-hoc networks is challenging
The average length (lc) of the AI code can be estimated by the
primarily due to the random movement patterns of various
following equation:
subscribers. Such estimation becomes comparatively easier in
C N D T (1)
Vehicular Ad-hoc Networks (VANETs) because the vehicles
where:
are confined to certain road maps [13]. In most of the cases, (i) C depends on the control information transmitted by
trains follow a similar route to the destination. This makes the network;
location estimation even easier. (ii) log
To effectively predict the status of the train under observation, (iii) log
we need a rough estimate of the location of the train as well as (iv) T depends on the type of system that is used.
the time. Therefore, if location information is available against The standard Unix time_t (data type that represents a point in
the timestamps, we can estimate the position of a train at a time) is a signed integer data type, traditionally of 32 bits.
particular time. Once location is known, the average time Therefore, it can traditionally represent 136 years uniquely. If
taken by the train to travel from that location to the such a system is followed, the minimum time that can be
destination can be estimated. Consequentially, we can represented is 1901-12-13, and the maximum time that can be
estimate the running status of the train. represented is 2038-01-19. This representation overflows at
To achieve the above, the CNDLE Model demands only 03:14:07 UTC 2038-01-19.
additions to the current cellular network architecture Therefore, in some newer operating systems, time_t has been
described in section – The addition of a Additional widened to 64 bits. Such a representation provides a unique
Information (AI) code and a location estimation technique. representation of 293 billion years and is sufficient for most
The AI code can contain additional information of a train such practical purposes.
as the train number, the distance covered by it and the Therefore, we can represent lc, the total number of bits required
timestamp. Once the mobile is successfully registered using to represent the AI code, by the following equation:
the MIN/ESN combination, the AI code can be transmitted to C log log 64 (2)
periodically as a control signal to update the running status of (Assuming 64 bit timestamps)
a particular train. The code on being received updates the
average time of the central database and the current status of
the train. B. LOCATION ESTIMATIONS
While a mobile is within its network, a mobile communicates via
control signals with its base station. The HLR maintains all
A. THE AI CODE administrative data that is related to each subscriber who is
In order to adequately represent the CNDLE model, we define registered in the corresponding GSM network, along with his
the AI code to contain the following: current location. However, the location provided by the HLR can
1. Train Number only help us identify the cell in which the user is active. A
2. The distance travelled by the train typical cell radius may vary from 4m – 30 km. Depending on the
3. Timestamp density mobile phones in an area, cells may be classified by their
A sample packet for AI code has been depicted in figure 2. radius:
 Macro-cellular: 1 - 30 km
 Micro-cellular: 200 - 2000 m
Control Train Distance  Pico-cellular: 4 - 200 m
Timestamp
Information Number Covered The Pico cells are characteristics of dense areas such as cities and
populated towns. In villages, Macro cells are prevalent.
Therefore, the mere knowledge of the location from the HLR can
C N D T
Figure 2 – Sample packet for the AI code
only provide us with a rough region in which a particular cell
phone is present.
Using figure 2 we can determine the average length of such a Moreover, the shape and extent of the cells keep changing with
packet. The following notions are used: time. A snapshot obtained by Siemens TORNADO D Cellular
dc: Current distance covered by the train Planning Tool of cellular structure is shown in figure 3.
dm: Maximum distance possible between two stations
dtot: The total distance to be covered during a journey.
nc: Train number of the current train
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 12, DECEMBER 2010, ISSN 2151‐9617 
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/  
WWW.JOURNALOFCOMPUTING.ORG       58 

A. PROCEDURES

Mentioned below are some procedures that need to be


implemented to predict the approximate arrival time of a
particular train (tarr) at a particular destination (dest) at a distance
ddest from the origin. The model proposes that the CRSS
communicates with the train via MS and BS. The packet Pack is
transmitted by CRSS on successful update of the information
provided by the train (via MS and BS). The packet Pwarn is
transmitted by the CRSS when more than one train is
Figure 3 - Figure depicting the actual cellular structure at a particular forbiddingly near on the same track. This communication
instant in time. happens via BS to MS that communicates the information to the
train transreceiver.
Therefore, for obtaining a more precise location, we can use the
information transmitted by the AI code. The AI code transmits dc ►Procedure to transmit train information to Mobile phones
which can be used for location estimation. Since trains generally nearby and fetch warning information. This procedure has to
take the same route to the destination, we can trace the distance implemented by the transmitting unit present in the train.
dc from its origin along the usual route of the train. The location
 
of the point which is distance dc away from the origin is the
PROCEDURE DIS_LOC(Packet Ki)
rough location of the train at time tc.
1. Check packet Ki. If Ki is Pack then the previous packet was
delivered properly and the system is ready for newer information.
If Ki is Pwarn, warn the driver, who is advised to slow down
because the same trace contains trains nearby which may cause
an accident.
2. Repeat steps 3-5 at regular intervals while dc < dtot
3. Update the distance covered (dc) from the origin
4. Form a packet containing dc, tc and nc.
5. Transmit the packet to nearby mobile phones.
6. Stop transmitting train information as the train has reached its
final destination
Figure 4 – Depiction of the origin (A), path followed (prominent line) and
the current location (B) by a train that has travelled distance dc ►Procedure facilitating formation and transmission of AI code to
nearby base station. This procedure has to implemented in
In figure 4 we assume that the train originates from point A and software on mobile stations.
reaches a point B after travelling a distance dc depicted by the  
dark line. Since the path of the train is known and the train PROCEDURE DIS_AI(Packet Pi)
follows a similar path during its journeys, we can estimate 1. Check for control information provided by the base station. If
location (latitude and longitude) based on the distance (dc) alone. Pi is Pack or Pwarn, forward it to the transreceiver in the train.
In the above figure we get the approximate location as 13° 56' N 2. Scan for the presence of signals transmitted by a train
and 75° 38' E. Therefore, if we know the distance travelled by a 3. If signals are present, execute steps 4-6 at whenever a packet
train, it becomes fairly easy to estimate the rough location of the Pi is received by the mobile
train. 4. Form the AI code from the information provided by the
packet Pi (dc, tc and nc) and control signals
5. THE CNDLE ALGORITHM 5. Transmit the AI code to the base station and delete
current packet Pi from memory
In this section we introduce the CNDLE algorithm for 6. Check for the presence for a new packet Pi. If present
estimation the current location of a particular train at a given return to step 3 else return to step 1.
moment of time. Different parts of the algorithm must be run 7. Goto step 1
at different locations to facilitate the proper functioning of the
model. We propose an algorithm for the determining the ►Procedure to receive and forward AI code transmitted by
estimated position of a train which may be used for predicting mobile phones in a train. This procedure has to be implemented
the running status as well as a safety application. Section A at the base station.
describes the procedures that are used in the algorithm.
Section B describes the use of the procedures for predicting PROCEDURE REC_AI(Packet Mi)
1. Check if packet Mi is Pack. If yes, forward the packet to the
the running status. Section C describes its use for accident
mobile station along with its control information.
avoidance.
2. Scan for the presence AI code in users control signals
3. If AI code is present, execute steps 4-6
4. Extract train information (dc, tc and nc) from the AI signal
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 12, DECEMBER 2010, ISSN 2151‐9617 
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/  
WWW.JOURNALOFCOMPUTING.ORG       59 
5. Transmit the information in form of a packet Ui to the make an estimation based in the average distance the
CRSS. train travels in the time tinit-tcur after it has travelled
6. Check for the presence for a new AI code. If present dst distance.
return to step 3 else return to step 1. 4. Compare the proximity of the trains. If the trains
7. Goto step 1 are forbiddingly close, repeat step 5 for all trains that
are near.
►Procedure to receive update current running status of trains. 5. Form a packet Pwarn and send it to the base
This procedure has to be implemented at the CRSS. station of the train in question.

PROCEDURE UPDATE(Packet Ui) B. PREDICTING RUNNING STATUS


1. Check for the receipt of packet Ui and repeat steps 2-14 if a
new packet Ui is received Based on the above procedures, the running status can be
2. Extract dc, tc and nc from the packet Ui predicted as mentioned by the procedure UPDATE(Packet Ui).
3. Edit the entry for the average time to travel distance dc by The arrival time at the location dest is provided by tarr and the
adding the new information to the existing database and delay can be predicted by tdel. No manual intervention is required
recalculating the average. in such a calculation. Since there is a central database which
4. Determine the rough location of the mobile using the maintains the train information, all destinations are updated
location mentioned in the HLR. accordingly.
5. Determine and store more precise location using dc by
tracing it along the usual route to determine the location in C. ACCIDENT AVOIDANCE
form of latitude (Plat) and longitude (Plong) (by the technique Once the procedure UPDATE(Packet Ui) is executed, we can
mentioned in section IV, B). estimate latitude (Plat) and longitude (Plong) of the train under
observation. When there are more than one trains on the same
►For any desired destination (dest) (for which we have to track, using the model, we can approximate the position of each
calculate the arrival time) at a distance ddest from the origin, we train at a given moment of time. Even if the location information
calculate the remaining distance and check the average time the of the train is stale due to the disconnections, we can effectively
train takes for travelling the distance to calculate the status predict the location of the train using the distance the train travels
6. if (ddest – dc >0) perform steps 7-9 on a average till a particular moment of time. Therefore, we have
7. Check the average time the train takes to reach a rough estimate of the location in terms of latitude and longitude
distance ddest (t2) and distance dc (t1). Now calculate the of all trains on the same track. Such information can be used to
average additional time (tavg) required to cover a prevent accidents by providing warning signals to trains. The
distance ddest from dc. procedure WARN_ALL() does this and transmits Pwarn packet to
Thus, tavg = t2 – t1 the BS which forwards it to the MS of the train and the MS in
8. Convert the timestamp tc received to time in hours turn forwards the Pwarn to the train transreceiver. Once the Pwarn
and minutes (tact) signal is received, the driver is warned and advised to slow down.
9. Calculate predicted time of arrival at destination dest Since the network connectivity provided by service providers in
(tarr) moving trains in intermittent and the location is a simple
tarr = tact + tavg estimate, this model will find only a limited use due to lack to
10. if (ddest – dc <0) the train has already passed ddest. accurate locations.
11. Calculate the delay in arrival (tdel) by calculating the
difference between the predicted arrival time (tarr) and the 5. THEORETICAL COMPARISON
actual arrival time (trt) .
This section provides the theoretical comparisons of the
tdel = tarr - trt
CNDLE model against the current model that is prevalent for
12. Update the delay for the particular destination (dest)
status inquiry used by Indian Railways. To keep track of trains
13. Provide an acknowledgement of the successful
in the present Indian scenario, the stations masters contact the
completion of the process by sending Pack to the base
control room at the divisional headquarters every time a train
station.
passes their station [12]. Until November 2009, the information
14. Check for receipt of a new packet Ui. If received, goto
step 2 so obtained used to be manually plotted on a map. Since then
15. Goto step 1 the mapping has been automated, but the backbone of the
system is still based on the calls of the station master. Such a
►Procedure to warn all the trains on the same track who are near system has two major problems: (1) the process is manual and
each other. This procedure has to be implemented at the CRSS. human intervention can introduce unnecessary delays. (2) The
information is updated only on stations. In case a train gets
PROCEDURE WARN_ALL() delayed between two stations, information cannot be relayed
1. Repeat steps 2-5 for all tracks that are active at that to successive stations. Tracking is therefore sometimes
moment of time. inaccurate.
2. Retrieve the current position of all trains on track The current model is based on the presence of a cellular
at that moment of time. network. As long as even a single user has a connection with
3. If the position provided is stale (say it is of time tst his base station, the information pertaining to the train’s
when distance travelled is dst) and current time is tcur, location can be transmitted with ease. This model is superior
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 12, DECEMBER 2010, ISSN 2151‐9617 
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/  
WWW.JOURNALOFCOMPUTING.ORG       60 
to the existing model because (1) The system is automated and [12] http://www.livemint.com/2010/06/06202515/Internal-conflict-
requires no manual intervention (2) the train’s position delays-satel.html?atype=tp. Retrieved 2010-8-1.
[13] J. Zhao and G. Cao,"VADD: Vehicle-assisted data delivery in vehicular
between stations can be estimated to a greater accuracy, as
ad hoc networks," in Proc. IEEE INFOCOM'06, Apr. 2006, pp. 1-12.
long as a cellular network is present. Avijit Gupta is currently pursuing the 5 year Integrated Post Graduate
programme at ABV- IIITM Gwalior. He secured a percentile of 99.85 in
GATE, 2009 (computer science) and is a regular participant of ACM-ICPC
6. CONCLUSION AND FUTURE WORK contests. His primary research areas are Computer networks and Adhoc
Networks.
In this paper, we presented the CNDLE Model for location
estimation for trains in Indian scenario. We described how the Shashikala Tapaswi received the B.E. degree from MITS, Gwalior, India,
in 1986, the M.Tech degree in computer science from University of Delhi,
current cellular network can be used to send the train running New Delhi, India, in 1993, and the Ph.D. degree in Computer Engineering
status using the AI code. The distance travelled by the train as from the Indian Institute of Technology, Roorkee, India, in 2002. Currently
well as the base station under which it resides at a moment she is a Professor at ABV-IIITM, Gwalior. Her primary research areas of
interest are AI, ANNs, Fuzzy logic, Digital Image Processing, Computer
provides us with a rough estimate of its location. The Networks, Mobile Networks, Adhoc Networks, information Security.
implementation of various procedures at different locations in
the network can eventually facilitate the transfer of
information using the cellular network.
For study and future work, the security aspects of such a
network can be discussed. Since the AI code is available to all
mobile stations, malicious manipulation of the code should
not be authenticated. Also, a study to reduce the packet length
of the proposed AI code can be discussed.

ACKNOWLEDGEMENT
We would like to acknowledge the support received by ABV-
Indian Institute of Information Technology and Management
Gwalior for facilitating this research.

REFERENCES
[1] Indian Railways Year Book (2006-2007). Ministry of Railways,
Government of India. 2007. pp. 2–3.
http://www.indianrailways.gov.in/deptts/stat-
eco/YearBook_06_07.htm. Retrieved 2010-8-3.
[2] Indian Railways Year Book (2006-2007). Ministry of Railways,
Government of India. 2007. p. 53.
http://www.indianrailways.gov.in/deptts/stat-
eco/YearBook_06_07.htm. Retrieved 2010-8-4
[3] Dinesh Bansal, “Wi-Fi based Train Safety & Communication Solution”,
Personal Wireless Communications, 2005. ICPWC 2005. 2005 IEEE
International Conference, pp 230-232, 2005.
[4] Eduard Bertran and José A. Delgado-Penín, “On the Use of GPS
Receivers in Railway Environments”, Vehicular Technology, IEEE
Transactions, pp 1452 – 1460, 2004
[5] Joseph Kee-Yin Ng, Junyang Zhou, Kenneth Man-Kin Chu and Karl R.
P, “A Train-Once Approach for Location Estimation Using the
Directional Propagation Model”, Vehicular Technology, IEEE
Transactions, pp 2242 – 2256, 2008
[6] H. Laitinen, S. Ahonen, S. Kyriazakos, J. Lahteenmaki, R. Menolascino,
and S. Parkkila, “Cellular location technology,” VTT Inf. Technol.,
Espoo, Finland, Tech. Rep. 007, Nov. 2001.
[7] P. Krishnamurthy and K. Pahlavan, “Principles of Wireless Networks:
A Unified Approach”, Upper Saddle River, NJ: Pearson Education,
2002.
[8] Lijie Chen and Tao Tang, “Research on the Influence of
Communication Delay on Safety Location of Train in Communication
Based Train Control (CBTC)”, Intelligent Vehicles Symposium, pp 3-5,
2009.
[9] Anup K. Ghosh, Vikram Rana and Barry W. Johnson, “A Distributed
Safety-Critical System for Real-Time Train Control Center for
Semicustom Integrated Systems”,
[10] Gary A. Bundell, “Aspects of the Safety Analysis of an On-board
Automatic Train Operation Supervisor”, Proceedings of the 2009 IEEE
international conference on Systems, Man and Cybernetics, pp 3223-
3230, 2009.
[11] http://tmrs.iitk.ac.in/tmrs/index.html. Retrieved 2010-8-1

También podría gustarte