Author: Dubey, Alka
Date published: July 1, 2010
Wireless Sensor Network (WSN) has recently come into prominence as the suitable solution platform for a number of sensing and monitoring applications. Advancements in Application Specific Integrated Circuits (ASIC) and Micro Electronic Mechanical Sensors (MEMS) have led to introduction of low cost and low power devices such as microcontrollers, radios and sensors. Such devices, which constitute the heart and soul of the sensor nodes, have boosted WSN usage. They leverage their deployment in large numbers to give robustness and fault tolerance with closer monitoring of events and environment. This way they are advantageous over the use of single higher capability sensors. Potential WSN deployment scenarios can be grouped under four application domains  namely, monitoring, inventory management, smart systems and military domain. All these application have the common theme of distributed sensing, low on-site computation and availability of low power sensing systems. WSN hardware consists of sensor nodes (motes), sensors, (if any) actuators, and at times higher end devices such as Intel's Stargate gateway  to connect sensor nodes to WiFi/ethernet LAN's. The software can be broken down into application specific, network management and OS related code. Newer designs have systems where the application or network management code and also a large chunk of OS code can be changed /upgraded on the fly [3-4]. Sensor nodes (motes) come in varied configurations based on the application for which they are to be used. They can be differentiated on number of parameters such as processing power, data bandwidth, battery power, etc. Thus they can be optimized on them based on application requirement. Generally, a sensor module consists of a microcontroller, a radio, a flash based memory, a power source (typically a battery), peripherals and connectors as shown in Fig. 1. Using the connector's one can connect a range of sensors to the mote. Since today's microcontrollers come with most of the functionalities such as ADC, DAC, UART, I2C, timers and program flash the number of peripheral chips are small in number leading to more compact designs. Some of the platforms also have onboard sensors for temperature, humidity or light. One can have external or internal radio antennae depending on the compactness of the design and the range of communication. The typical sensors used in conjunction with low powered sensor networks are temperature, pressure, light, humidity, acceleration, velocity, displacement, sound etc.
Indian Railways operate on gigantic dimensions covering over 63,000 route kilometers with daily loading of 1.6 million tones of freight and daily transporting of 14 million passengers by logging more than 2 million train kilometers per day. Safety is of paramount importance to Indian railways. Highest priority is accorded to safety and the rail mode in India continues to be the safest means of transportation for public. No compromise is tolerated in Safety of Rail users and all levels of management keep reviewing the Safety performance from time to time [5-6].
In India 38341 level crossings are present in which 16549 level crossing are manned and rest 21792 level crossings are unmanned. On this network 14 million people and more than a million tones of freight moves daily [6, 31]. Thus it is essential that all the aspects of this network are up and running securely and safely round the clock. If all unmanned level crossings are to be manned, Railways require approximately Rs. 2450 crores as Capital cost to man them and approximately Rs. 700 crores per annum will be required to meet the maintenance and operation cost. The cost of manning with interlocked signals will be around Rs. 5500 crores. However, to eliminate probability of any accident at manned and unmanned level crossings, construction of Road over bridges and Road under bridges may be envisaged, but it will involve staggering amount of Rs. 4,00,000 crores. There 20 % accidents occur due to manned and unmanned level crossing. Thus it becomes essential to monitor the level crossing of Indian railways for enhancing security of people and make it more reliable. Fig. 2 shows the analysis of level crossing accidents and fatalities .
For monitoring of level crossing we provide a viable solution, which is scalable, portable and easier to install without requiring any on-site skilled support. The proposed architecture gives on-demand data retrieval capabilities and can be installed at remote sites with maintenance cycles spanning months. This solution uses relatively inexpensive MEMS accelerometers along with wireless sensor network (WSN) to achieve the goal.
In the present work a wireless sensor network is developed which monitor the level crossing of Indian railways WCR zone and control the gate operation as well as provide signal to pilot and co pilot about the track status.
2. Sensor Node Architecture
A wireless sensor node [8-9] is a small, low power, ideally inexpensive device which can be deployed in large numbers. Each node has a limited processing power, limited memory and battery supply as well as limited range of communication and region for which sensor data has been collected. A few generic and commercial sensor nodes are shown in Fig. (3). A collection of such nodes co-ordinating amongst them constitutes a wireless sensor network (WSN). A typical WSN consists of sensor nodes (motes), corresponding sensors, and software to run and maintain the node and the network. Fig. 4 gives the architecture of a sensor node with different subsystems. A general-purpose sensor node has a microcontroller for computation and control, a radio for communication, a power source (usually a battery pack) and a set of interfaces/ports to connect to sensors, actuators or other application specific electronics or auxiliary circuits. These nodes are programmed either by connecting them to a computer or a special programming board. It is not uncommon to have certain generic sensors such as temperature, humidity and photo sensors located on the motes. Depending on type of application and cost involved it can also provide additional data memory (such as flash), analog to digital and/or digital to analog converters. The software on the mote can be application specific code or generic program with deployable modules to change the functioning of the nodes.
3. Operational Goal
The train (engine) contains a powerful radio (IEEE 802.11), and transmits beacons continuously as it arrives near the level crossing. The sensor mote (mote A), which is kept at a distance of d2 from the level crossing gate in the direction of the arrival of the train, wakes up on sensing the channel on which the train is transmitting beacons using Wake-on- WLan technology .
This mote does a duty cycling between sleep and wake-up. This means that its sleep and wake up time have to be such that it positively detects the arrival of the train [8-9]. Now mote A wakes up mote B (base node) by transmitting 802.15.4 packets, which in turn wakes up other sensor nodes similarly by transmitting wakeup packets. The wake-up mechanism is depicted in Fig. (5). The sensor mote connected to the soekris (base node) is responsible for powering on the soekris board as well as the other nodes deployed on the piers of the level crossing. These sensor nodes collect vibration data from the tri-axial accelerometer and transmit it to the base node through the multi-hop wireless sensor network. The base node (sensor node) transfers this data to the soekris. The soekris transmits data pertaining to previous passage of the train to soekris sitting on the train. The data collected at this point of time is transmitted to the next train passing by. The train, which has collected the data from the sensor network ultimately, feeds the backbone at any of the stopping stations. The data is either available online or made available for further processing.
4. Data Organization
The solution employs a Wireless Sensor Network (WSN), each of the nodes in the WSN connected with an accelerometer. Every node collects vibration data (accelerometer readings) as the train passes on the track of level crossing. The setup (accelerometer and sensor network) is responsible for measuring the train introduced ground vibration. It is important to know the exact amount of data that each node will collect. Table (1) shows the calculation of amount of data at each node.
The amount of time the train remains on the level crossing is calculated by equation (1).
T = (L+T)/[(V*1000)/3600] seconds (1)
Thus the time taken by train for crossing level crossing area is between 9 minutes to 15 minutes.
The total amount of data is calculated by equation (2)
D = r*a*b*t (2)
Total amount of data is 10.25 kB to 13.25 kB. In India, trains run at an average speed of 80-140 km/h (fastest trains), but their speed is often lesser when they pass on some level crossing area, so we have taken the value 20-60 km/h. Generally the level crossing area is equal to the width of road but we have taken 5 KM more area at both side of level crossing including the area of level crossing. Thus 10 km area is covered in the form of level crossing. The length of a train is around 500 m (30 m per coach and 20 coaches). We need to sample at a rate of 10 samples for second to get useful data about the vibration is necessary.
The data collected in each node is around 10.25 kB to 13.25 kB, but the tmote has only 10 kB RAM. But the RAM is also used for storing other local variables and arrays, so the effective RAM available for actual data storage will be around 6-7 kB (in our present implementation). So we store 10.25 to 13.25 kB of data in tmote in the huge flash that the tmote features. We create an abstraction called a data File. A data File is an abstraction created to handle the amount of data generated at a single node, which is greater than the amount of RAM available. The size of the File is fixed in a network. All data generated at a node is divided into fixed number of files. The number of files at all nodes is the same because the amount of data generated for an event at every node is the same. The data File has the following attributes.
* The node id of the node, which originally collected the data.
* The file no within that node.
* Start address of the file in the flash.
* Size of the file (for convenience we keep the size of the file constant, this does not affect the performance) .
We divide all the data node collects into files; the size of the files is small enough that files can be easily stored in the RAM. But these files need to be transferred to the flash before the arrival of data for the next file. The files are transferred one file at a time to the next node (parent node) in the network. The transferred file is stored in the flash before the next file is transferred. So in this way we store and transfer data in the form of files. But we store all the properties of the File in the RAM. Once the data is stored on the flash as files, data is also read as files and transmitted to the lower layer (transport) as a buffer to be reliably transmitted to the next node in the network as packets, where the packets are then reassembled and reconstructed as a buffer in the transport layer, which is then passed as a file to the application layer, where it is again stored as a file in the flash. This node is now responsible for transmitting not only its file but also the file that it has received from its children. Consider a 10 node network arranged as a temple topology with the base node and nodes numbered 1-9. Consider that the entire data at each node is divided.
In this work we present a mechanism for a level crossing Monitoring system of Indian railways. The proposed system has a number of advantages over the currently used wired systems. Our system is easier to deploy, lower in cost and autonomous in its operation. It can control the operation of level crossing gates automatically and also provide data on pilot screen at engine, which provide the status of the track with real time. The system is designed such that it can be left on site and the data can be retrieved on demand. We propose a new approach to fetch the data from the remote site using the train itself. We also use a new approach of Wake-on- WLAN to detect 802.11 signals with 802.15.4 compliant radios. This method is used to detect the approaching train, which gets used as a vibration generation source (similar to mechanical shaker) for the current run and as a data transporter for data collected at an earlier round.
. M. Marti, B. Kusy, G. Simon and. Ldeczi, The flooding time synchronization protocol, in Proc. of the SenSys '04, Nov. 2004.
. Intel's Stargate systems. http://www.xbow.com/Products/productsdetails. aspx?sid=85
. Deluge: TinyOS Network Programming, http://www.cs.berkeley.edu/~jwhui/research/deluge
. P. Levis, D. Gay, and D. Culler, Active Sensor Networks, In NSDI'05, May 2005.
. White Paper - Safety on Indian Railways, Govt, of India, Ministry of Railways, April 2006.
. Alka Dubey et al, design & development of black box for railway electric engine WAP-7, in Proc. of the 4th M. P. Science Congress, 2007, pp. 40-49.
. Amitabh, Indian Railways - Safety Performance of Relevant Years, Published by Safety Directorate of ministry of Railways, Govt, of India.
. Nilesh Mishra. Brimon, Application design for railway bridge monitoring, Master's Thesis, IIT Kanpur, 2006.
. Nilesh Mishra, Kameswari Chebrolu, Bhaskaran Raman, and Abhinav Pathak, Wakeonwlan, in Proc. of the 15th International World Wide Web Conference (WWW '06), 2006.
.Hemanth Haridas, Brimon, Application design for railway bridge monitoring, Master's Thesis, IIT Kanpur, 2006.
2010 Copyright ©, International Frequency Sensor Association (IFSA). All rights reserved.
Department of Physics & Electronics
Dr. Harisingh Gour Central University, Sagar (M.P.), India
Received: 23 July 2010 /Accepted: 30 July 2010 /Published: 31 July 2010