Wireless sensor networks (WSN). Wireless sensor network

I want to devote my article to the technologies of wireless sensor networks, which, it seems to me, are undeservedly deprived of the attention of the habra community. I see the main reason for this is that the technology has not yet become mass and for the most part is more interesting to academic circles. But I think in the near future we will see many products based in one way or another on the technologies of such networks. I have been researching sensor networks for several years, wrote a PhD thesis on this topic and a number of articles in Russian and foreign journals. I also developed a course on wireless sensor networks, which I read in the Nizhny Novgorod State University(I don’t provide a link to the course, if you are interested, I can give a link privately). Having experience in this field, I want to share it with a respected community, I hope you will be interested.

General information

Wireless sensor networks have received a lot of development in recent years. Such networks, consisting of many miniature nodes equipped with a low-power transceiver, microprocessor and sensor, can link together global computer networks and the physical world. The concept of wireless sensor networks has attracted the attention of many scientists, research institutes and commercial organizations, which has provided a large flow of scientific papers on this topic. Great interest in the study of such systems is due to the wide possibilities of using sensor networks. Wireless sensor networks, in particular, can be used to predict equipment failure in aerospace systems and building automation. Due to their ability to self-organize, autonomy and high fault tolerance, such networks are actively used in security systems and military applications. The successful application of wireless sensor networks in medicine for health monitoring is associated with the development of biological sensors compatible with integrated circuit sensor nodes. But wireless sensor networks are most widely used in the field of monitoring the environment and living beings.

Iron

Due to the lack of clear standardization in sensor networks, there are several different platforms. All platforms meet the basic basic requirements for sensor networks: low power consumption, long operating time, low-power transceivers and the presence of sensors. The main platforms include MicaZ, TelosB, Intel Mote 2.

MicaZ

Microprocessor: Atmel ATmega128L
7.3728 MHz frequency
128 KB flash memory for programs
4 KB SRAM for data
2 UART's
SPI bus
I2C bus
Radio: ChipCon CC2420
External flash memory: 512 KB
51-pin auxiliary connector
eight 10-bit analog I/Os
21 digital I/Os
Three programmable LEDs
JTAG port
Powered by two AA batteries

TelosB

Microprocessor: MSP430 F1611
8 MHz frequency
48 KB flash memory for programs
10 KB RAM for data
SPI bus
Built-in 12-bit ADC/DAC
DMA controller
Radio: ChipCon CC2420
External flash memory: 1024 KB
16-pin additional connector
Three programmable LEDs
JTAG port
Optional: Light, humidity, temperature sensors.
Powered by two AA batteries

Intel Mote 2

320/416/520 MHz PXA271 XScale microprocessor
32 MB Flash
32 MB RAM
Mini USB interface
I-Mote2 connector for external devices (31+21 pin)
Radio: ChipCon CC2420
LED indicators
Powered by three AAA batteries

Each platform is interesting in its own way and has its own characteristics. Personally, I had experience with the TelosB and Intel Mote 2 platforms. Also, our own platform was developed in our laboratory, but it is commercial and I cannot talk about it in detail.

The most common 3 years ago was the use of the CC2420 chipset as a low power transceiver.

Software and data transfer

The main standard for data transmission in sensor networks is IEC802.15.4, which was specifically designed for wireless networks with low power transceivers.

There are no standards in the field of software in sensor networks. There are several hundred different protocols for processing and transmitting data, as well as node management systems. The most common operating system is an open source system - TinyOs (while at Stanford University, I personally met one of the developers). Many developers (especially commercial systems) write their own control system, often in Java.

Sensor node control program running operating system TinyOs is written in nesC.

It should be noted that due to the high cost of equipment and the complexity of setting up sensor networks, various simulation systems have become widespread, in particular the TOSSIM system, specially designed to simulate the operation of nodes running TinyOs.

Conclusion

Sensor networks are becoming more widespread in Russia. When I started doing them in 2003, the number of people in Russia who were familiar with this technology could be counted on the fingers. Including in Russia, the notorious Luxsoft Labs was engaged in this.

I have worked with sensor networks for 6 years and I can tell you a lot about these technologies. If the Habrasociety is interested and I have the opportunity, then I will be happy to write a series of articles on this topic. I can touch on such things as: real work with the TmoteSky platform, programming features for the TinyOs system in nesC, original research results obtained in our laboratory, impressions of 1.5 months of work at Stanford University, in a project on sensor networks.

Thank you all for your attention, I will be happy to answer your questions.

The analysis of wireless sensor networks has been carried out. The Omnet++ program was chosen for research. Research task implemented model range wireless sensor networks and evaluation of the parameters of their functioning. The following tasks were solved: the model of energy consumption in wireless sensor networks was improved, an algorithm for the operation of this model was proposed, which makes it possible to reduce delays in the transmission of packets between nodes. Developed computer model in the chosen program, it is proved that the application of this model is effective and expedient in practice. In this article, a study was made of the energy consumption of network nodes. It is energy consumption that is the key parameter of the quality of functioning of wireless sensor networks, so the question of its calculation when creating such systems arises first. Work carried out detailed analysis energy consumption of nodes of wireless sensor networks, as well as a method for calculating the energy consumption of end nodes. Various approaches have been proposed to reduce energy consumption. The main point of energy efficient network operations will be the ability to put more nodes into sleep mode, directly to increase their battery life. Also, in sensor networks using ZigBee technology, it is possible to compress information before sending it. The amount of energy consumed will similarly depend on the selected network topology. It has been proven that the lowest energy costs occur when using star or cluster tree topologies, because in these topologies the coordinators are directly connected to the fixed network.

wireless sensor network

Omnet++ program

transmission delays

power consumption of sensors

network bandwidth

energy saving

1. Terentiev M.N. The method of functioning of systems for monitoring the parameters of objects with a changeable configuration based on discrete wireless sensor networks: dis. … cand. tech. Sciences: 05.13.15 / M.N. Terentiev. - Moscow, 2010. - 154 p.

2. Khusnullin V.I. Study of energy consumption of nodes in a wireless sensor network / V.I. Khusnullin, E.V. Glushak // Proceedings. report at the II Scientific Forum "Telecommunications: Theory and Technologies (TTT)" at the XVIII International scientific and technical conference"Problems of engineering and technology of telecommunications". - Kazan, 2017. - T. 2. - S. 10–13.

3. Ivanova I.A. Determination of the perimeter of the coverage area of wireless sensor networks / I.A. Ivanova // Industrial ACS and controllers. - 2010. - No. 10. - P. 25–30.

4. Vlasova V.A. Analysis of energy cycles of nodes of wireless sensor networks / V.A. Vlasova, A.N. Zelenin // Eastern European Journal of Advanced Technologies. - 2012. - V. 3, No. 9 (57). – P. 13–17.

5. Galkin P.V. Features of the implementation of wireless sensor networks based on ZigBee technology: mater. VI intl. scientific-practical. conf. / P.V. Galkin, D.V. Karlovsky // Actual problems of sciences. - 2010. - No. 31. - P. 7–11.

6. Baskakov S. Evaluation of power consumption of wireless nodes in MeshLogic networks / S. Baskakov // Wireless technologies. - 2010. - No. 1. - S. 28–31.

7. Kireev A.O. Distributed system for energy monitoring of wireless sensor networks / A.O. Kireev, A.V. Svetlov // Izvestiya SFedU. Technical science. - 2011. - No. 5 (118). – P. 60–65.

8. Daniel Kifetew Shenkutie, Residual Energy Monitoring in Wireless Sensor Networks / School of Information Science, Computer and Electrical Engineering Halmstad University. - 2011. - 84 p.

9. Kramorenko E.G. Reducing the power consumption of sensor networks due to preliminary data compression: mater. to IV All-Ukr. sci.-tech. conf. / E.G. Kramorenko, M.V. Privalov // Information control systems and computer monitoring 2013. - Donetsk: DonNTU, 2013. - P. 364–369.

Recent advances in semiconductor, networking, and logistical technologies are driving widespread deployment of large-scale wireless sensor networks (WSNs).

A wireless sensor network is a distributed, self-organizing network of many sensors (sensors) and actuators, interconnected via a radio channel. Moreover, the coverage area of such a network can range from several meters to several kilometers due to the ability to relay messages from one element to another.

A wireless sensor network model was proposed. To evaluate the effectiveness of the proposed model, we will perform the simulation in software package OMNet++. Let us analyze the simulation procedure and simulation results. OMNeT++ is an object-oriented network simulator with discrete event.

There are two types of packets in the simulation: message packets, which are used by sensor nodes in the network to send information to the receiver node, and the second type is the energy packet, which is used to transmit energy information to the monitoring node. In the simulation, each node periodically calculates the amount of energy consumed and also predicts the amount of energy it will consume in the upcoming period. The amount of energy consumed is compared with the predicted one: if the difference between them is greater than a certain threshold, the node will send an energy packet to the main network node (base station). Some of the packages contain information about the predicted energy consumption in the nodes. The numerical values chosen for the simulation can be seen in the table below.

Numerical values used

These values are used in all simulations. To demonstrate the effectiveness of the proposed prediction model, a network with a hundred nodes is implemented. The nodes on the network use a chosen routing protocol, called MFR, to forward the packet to the destination node. A node using MFR forwards data to a node in its transmission range.

On fig. 1 node S transmits its data to node M because it is closer to receiver D than other nodes in its transmission range when it is projected onto the line connecting node S and receiver D. Sensor nodes use a system position message to notify their the location of their neighbors. The sensor nodes in the network populate the routing table with the location of their neighbors, and choose the nearest one as the next one to transmit data.

Let's imagine simulations performed using OMNeT++ simulation. The error between the residual energy in each node and the value recorded in the monitoring node for a different threshold value is analyzed. Next, the relationship between the number of power packets sent to the control node and the threshold used is examined. Energy cost is the energy expended by the nodes in the network to store information in the control node regarding the amount of residual energy left in their batteries. This network energy depends on the average number of energy packets sent to the monitoring node by each sensor node. On fig. 2 shows the average number of packets sent per node for different thresholds when E = 100 s.

After running the simulation for two and a half hours, the simulation results are shown in Fig. 2 and 3. The graphs in the figures show the number of power packets sent to the control node for three prediction periods (T = 200, T = 300 and T = 400) when two different maximum event arrival rates (E = 100 s and E = 50 from). The graphs in the two figures showed how the rate of arrival increases, the number of energy packets sent usually increases. For the same data arrival rate, the number of energy packets sent increases as the prediction error threshold decreases.

Rice. 2. Average number of packets sent per node when E = 100 s

Rice. 3. Average number of packets sent to the node when E = 50 s

On fig. 4 and 5 show the number of energy packets sent at the occurrence of an event that trigger the sensor node sensors are considered strictly periodic. Used arrival periods between events P = 50 and P = 100 s. According to the graphs, the number of energy packets sent from each node has increased as the arrival time of the event has decreased. Over the same period, the number of packets sent increased as the threshold decreased.

Energy when building an energy map is directly related to the amount of energy spent, as a result of which it also increases, since the prediction error threshold is reduced. The results of the simulations performed also showed that the prediction period increases the number of energy packets sent. This is because, with longer prediction intervals, the energy consumption of nodes exhibits a more periodic pattern than shorter prediction intervals. This results in a more accurate power consumption prediction because the method relies on the nodes' past power consumption history for prediction.

Rice. 4. Average number of packets sent to the node when P = 100 s

Rice. 5. Average number of packets sent to the node when P = 50 s

On fig. 6 shows a comparison of the results obtained with the exponential use of the method proposed in this paper and the results found in . The comparison is made based on the average number of energy packets sent to the monitoring node for various thresholds.

Typically, the average energy of packets sent to the monitoring node is higher for all thresholds used when the exponential model is used than the probabilistic method in when it is assumed that the occurrence of events in the environment is uniformly distributed. This is because the exponential averaging method predicts the upcoming power consumption of nodes based on their power consumption history. Due to the occurrence of unexpected events, the behavior of some of the energy-consuming nodes may deviate from the average energy that they used in the past. This affects the expected future power depletion predictions of the nodes, prompting the nodes to send large quantity packages.

Rice. 6. Comparison of models (average number of packets sent per node)

The higher the number of energy packets sent to monitor the node, the higher the cost of building an energy map. In the case of a strictly periodic event arrival model, the exponential model used in this work shows better performance than the model used in when the threshold is set to 1% and 3%. This is due to the constant energy consumption of nodes associated with the periodic nature of events.

On fig. 7 and 8 show the total number of packets in the network for two different models arrival of packages. In both cases, the total number of energy packets in the network increases as the threshold value decreases, while the number of message packets remains unchanged. An increase in the total number of energy packets increases the cost of the energy card, since it is directly related to the number of energy packets sent from the sensor node. Both figures show the total number of packets in the network for the entire simulation period when the prediction period is set to 400 s.

The energy monitoring score is the difference between the residual energy of each node and the residual energy recorded at the control node. As a result of the evaluation, we conclude that the amount of energy exceeding the threshold value accumulates in the monitoring node and the deviation is greater for higher threshold values.

1) The main point of energy efficient network operations will be the ability to put more nodes into sleep mode, directly to increase their battery life. When the sensor node is active, it can go into sleep mode, allowing it to reduce power consumption. The sensor node switches to this mode between sessions of receiving/transmitting data. All modes consist of cycles, and each cycle will consist of periods of sleep and periods of listening. The maximum energy costs will be in the transmission and reception of data. Namely, one of the options for reducing power consumption would be to switch the sensor from active mode to sleep mode, when power consumption is minimal;

2) in sensor networks using ZigBee technology, it is possible to compress information before sending it. This reduces the time of data transmission, the device itself reduces the time of its stay on the air and, of course, consumes less energy to transmit a data packet. Codecs are required for direct compression. The use of codecs allows you to reduce energy consumption by compressing the transmitted information. Minimizing the amount of data being broadcast will result in lower power consumption.

3) the amount of energy consumed will similarly depend on the selected network topology. Energy is spent more in a cell topology due to the fact that each network node communicates more often, and, therefore, it is more in working order.

Rice. 7. Total number of packets in the network for P = 50

Rice. 8. Total number of packets in the network for E = 50

The lowest energy costs occur when star or cluster tree topologies are used, because in these topologies the coordinators are directly connected to the fixed network.

Bibliographic link

Achilova I.I., Glushak E.V. RESEARCH OF WIRELESS SENSOR NETWORKS // International magazine applied and fundamental research. - 2018. - No. 5-1. - P. 11-17;
URL: https://applied-research.ru/ru/article/view?id=12208 (date of access: 04/26/2019). We bring to your attention the journals published by the publishing house "Academy of Natural History"

Wireless sensor networks have unique characteristics of easy deployment, self-organization and fault tolerance. Emerging as a new information gathering paradigm, wireless sensor networks have been used for a wide range of health, environmental, energy, food safety and manufacturing applications.

Over the past few years, there have been many prerequisites for sensor networks to become real. Several sensor node prototypes have been created, including Motes at Berkeley, uAMPS at MIT (at the Massachusetts Institute of Technology), and GNOMES at Rice. The elementary functions of sensor networks are positioning, detection, tracking, and detection. In addition to military applications, there have also been civilian applications based on elementary functions, which can be divided into environmental control, environmental monitoring, health care and other commercial

applications. In addition, Sibley recently created a mobile sensor called Robomote, which is equipped with wheels and is able to move around the field.

As one of the first attempts to use sensor networks for civilian applications, Berkeley and Intel Research Laboratory used the Mote sensor network to monitor storm readings in the Great Duck Islands, Maine in the summer of 2002. Two-thirds of the sensors were installed off the coast of Maine to collect the necessary (useful) information in real time on the world wide web (Internet). The system worked for more than 4 months and supplied data

within 2 months after the scientists left the island due to bad weather conditions (in winter). This habitat monitoring application is an important class of sensor network applications. Most importantly, network sensors are able to collect information in dangerous environments that are unfavorable to people. In the course of monitoring studies, design criteria were considered, including design, creation, creation of a sensor system with the possibility of remote access and data management. Numerous attempts have been made to achieve the requirements, leading to the development of a set of prototype sensor network systems. The sensor system used by the Berkeley and Intel Research Laboratory, although primitive, was effective in collecting interesting environmental data and provided scientists with important information.

Sensor networks have found applications in the field of observation and prediction (guessing). A living example of such an application is the Automated Local Evaluation in Real-Time (ALERT) system developed by the National Weather Service with a wireless network of sensors. Equipped with meteorological/hydrological sensor devices, sensors in these conditions usually measure several properties of local weather, such as water level, temperature, wind. The data is transmitted via a direct radio link (line-of-sight radio communication) via sensors at the base station. The Flood Forecast Model has been adapted to process the data and issue automatic warnings. The system provides critical real-time rainfall and water level information for assessing potential flooding anywhere in the country. The present (current) ALERT system is installed throughout the US West Coast and is used for flood warning in California and Arizona.

Recently, sensor systems have been used extensively in the healthcare industry, used by patients and physicians to track and monitor glucose levels, cancer detectors, and even artificial organs. Scientists suggest the possibility of implanting biomedical sensors into the human body for various purposes. These sensors transmit information to an external computer system via the wireless interface. Several biomedical sensors are combined into a system of applications to determine the diagnosis and treatment of the disease. Biomedical sensors herald a more advanced level of medical care.

The main difference between wireless sensor networks and traditional computer and telephone networks is the absence of a permanent infrastructure that belongs to a particular operator or provider. Each user terminal in the sensor network has the ability to function not only as an end device, but also as a transit node, as shown in Figure 1.2.

Figure 1.2 - An example of connecting network sensors

I want to devote my article to the technologies of wireless sensor networks, which, it seems to me, are undeservedly deprived of the attention of the habra community. I see the main reason for this is that the technology has not yet become mass and for the most part is more interesting to academic circles. But I think in the near future we will see many products based in one way or another on the technologies of such networks. I have been researching sensor networks for several years, wrote a PhD thesis on this topic and a number of articles in Russian and foreign journals. I also developed a course on wireless sensor networks, which I read at the Nizhny Novgorod State University (I don’t give a link to the course, if you are interested, I can give a link privately). Having experience in this field, I want to share it with a respected community, I hope you will be interested.

General information

Wireless sensor networks have received a lot of development in recent years. Such networks, consisting of many miniature nodes equipped with a low-power transceiver, microprocessor and sensor, can link global computer networks and the physical world together. The concept of wireless sensor networks has attracted the attention of many scientists, research institutes and commercial organizations, which has provided a large flow of scientific papers on this topic. Great interest in the study of such systems is due to the wide possibilities of using sensor networks. Wireless sensor networks, in particular, can be used to predict equipment failure in aerospace systems and building automation. Due to their ability to self-organize, autonomy and high fault tolerance, such networks are actively used in security systems and military applications. The successful application of wireless sensor networks in medicine for health monitoring is associated with the development of biological sensors compatible with integrated circuit sensor nodes. But wireless sensor networks are most widely used in the field of monitoring the environment and living beings.

Iron

MicaZ

Microprocessor: Atmel ATmega128L
7.3728 MHz frequency
128 KB flash memory for programs
4 KB SRAM for data
2 UART's
SPI bus
I2C bus
Radio: ChipCon CC2420
External flash memory: 512 KB
51-pin auxiliary connector
eight 10-bit analog I/Os
21 digital I/Os
Three programmable LEDs
JTAG port
Powered by two AA batteries

TelosB

Microprocessor: MSP430 F1611
8 MHz frequency
48 KB flash memory for programs
10 KB RAM for data
SPI bus
Built-in 12-bit ADC/DAC
DMA controller
Radio: ChipCon CC2420
External flash memory: 1024 KB
16-pin additional connector
Three programmable LEDs
JTAG port
Optional: Light, humidity, temperature sensors.
Powered by two AA batteries

Intel Mote 2

320/416/520 MHz PXA271 XScale microprocessor
32 MB Flash
32 MB RAM
Mini USB interface
I-Mote2 connector for external devices (31+21 pin)
Radio: ChipCon CC2420
LED indicators
Powered by three AAA batteries

The most common 3 years ago was the use of the CC2420 chipset as a low power transceiver.

Software and data transfer

The main standard for data transmission in sensor networks is IEE802.15.4, which was specifically designed for wireless networks with low-power transceivers.

The control program for the sensor node under the control of the TinyOs operating system is written in the nesC language.