Navigation  
RU| EN| UK
Artem Timkov
Master of Donetsk National Technical University Artem Timkov
Faculty: Computer Science and Technology
Department: Automated Control Systems
Speciality: Information Control Systems and Technologies
Theme of Master's Work: Optimization of Wireless Sensor Systems Using Ant Algorithms
Scientific Supervisor: Ph. D.(inEngineering), Associate Professor of ACS Alexander O. Telyatnikov
Abstract

of the qualification master’s work

"Optimization of Wireless Sensor Systems Using Ant Algorithms"

Nowadays, science stride forward. People have developed microcomputers, microprocessors, have created a robot that can analyze the information obtained from outside, etc. Have been developed new algorithms, methods of solving various problems.

One such new technology is – “Smart Dust. “Smart Dust” - is a technology for wireless sensor networks. She began to develop in the mid 90's. The most successful research in this area was research professor at the University of California, Christopher Piester. After that, these studies have been actively used defense research center USA (DARPA). Originally designed for military use, sensor networks have been actively used for civilian purposes. To date, the market of wireless sensors is actively developing, developed the ZigBee Alliance for industrial support of new technology, which includes more than 100 companies.

This technology is so universal that it can be used practically in all spheres of human activity. Due to the wide range of sensors can monitor all sorts of indicators (temperature, sound, motion, pressure, etc.). Ability to pass from one node to another allows you to deploy a network for tens of thousands of square kilometers. For example, you can monitor the vital parameters of severely ill in hospital, while staying in one place, or to permit the transfer of data on changes in temperature or ignition in the woods. And such examples are a huge number.

The main task in developing such projects is to ensure reliable communications, fast delivery of data packets to the minimization of energy. To date, the issues related to the creation and deployment of sensor networks is the subject of many works, mostly by foreign authors. The subjects of works ranges from highly specialized issues relating to the creation of a separate component of the network facilities (transceivers, microcontrollers, sensors, etc.) with low price and low power consumption to the problems that arise in the operation of wireless sensor networks (questions related to the organization of the network, development software, the binding location of objects on the network to geographic coordinates, etc.).

Relevance of the topic of master due to several factors:
- Firstly, for the moment there are no common standards for the design of wireless sensor networks. Each task requires a special approach to the choice of network topology, algorithms, data, methods of processing, etc.;
- Secondly, the current development designed to build networks of very large sizes, which are the most demanding of optimal settings of the system.

Qualifying work performed during the Masters 2009-2010 years. in accordance with the scientific direction of the Department of "Automated Control Systems", Donetsk National Technical University.

The purpose of my master's thesis is to minimize the cost of energy nodes wireless sensor network. To achieve this goal have been made following tasks:
     1. Analysis of existing systems modeling wireless sensor systems and choose the most suitable system.
     2. Development of a simulation model wireless sensor networks.
     3. An analysis of existing data transmission protocols in wireless sensor networks.
     4. Defining a benchmark for comparison of protocols.
     5. Analysis of parameters affecting the minimization of energy.
     6. Development of algorithm data based on ant algorithms.
     7. Development of a new data transfer protocol.
     8. Testing and comparison of new data transfer protocol with other protocols.

The novelty of the master's thesis is as follows: development of a new data transfer protocol, which enables to minimize power consumption when transmitting data between nodes by finding the optimal route.

Nowadays, developments in wireless sensor networks are being given the large spatial coverage. The new data transfer protocol will choose the best routes data that will provide a more efficient network and sensor nodes.

Wireless sensor networks

Wireless sensor network (WSN) - a wireless system, which is a distributed, self-organizing and robust to failures of individual elements of a network whose nodes are special devices - prodigals. WSN - a wireless network, transmission of information in which is produced from one node to another, until the packet reaches the remote gateway. From the gateway information comes to parent computer, which performs information processing, or passes it on.

Mote is a dress size is usually not more than one cubic inch. The board placed the processor, memory - flash and operational, analog and analog-digital converters, radio frequency transceiver, power supply and sensors. Sensors can be very diverse, and they are connected via both digital and analog connectors. Most widely used temperature sensors, pressure, humidity, light, vibration, less often - magneto, chemical (eg, measuring the content of CO, CO2), audio, and some others [1]. Set sensors used depends on the functions performed by wireless sensor networks. Nutrition Mota made from a small battery. Prodigals used only for collection, primary processing and transmission of sensory data. Appearance Motov produced by various manufacturers, shown in Fig. 6.1

Figure 6.1 – Mote appearance.
Figure 6.1 – Mote appearance.

The main functional processing of data collected motami, carried out on the node, or gateway, which is powerful enough kompyuter.Problema receiving sensory information gathered motami, is solved as follows. Prodigals can share among themselves information using transceivers operating at radio frequencies. This is, first, sensory information is read from sensors, and secondly, information on the status of devices and the results of the data. Information is transferred from one Motov other chain, and eventually coming to the Gateway prodigals throw him the accumulated information. If part of Motov fails, the work of a sensor network after reconfiguration must continue. But in this case, of course, decreases the number of information sources [2].

The benefits of technology of wireless sensor networks can be effectively used for various applications related to distributed collection, analysis and transmission of information [3]. Wireless sensor networks - a new promising technology, and all associated projects are mainly in the development stage. Describe the main areas of application of this technology:
     - Defense and security;
     - Control environment;
     - Monitoring of industrial equipment;
     - Security systems;
     - Monitoring the status of agricultural land;
     - Energy management;
     - Control of ventilation, air conditioning and lighting;
     - Fire alarm;
     - Stock control;
     - Monitoring the transport of goods;
     - Monitoring the physiological state of man;
     - Control staff, etc.

The presented list is extremely small and virtually unlimited. From the rather large number of examples of the use of wireless sensor networks distinguish the two.

Wine industry - occupation, requiring the myriad variety of information. The wine is bought and sold not by weight or volume (as, for example, cereals or vegetables), and its price depends largely on the location and year of cultivation and harvest. The quality of wine is influenced by many factors, and experienced winemakers carefully takes into account the tiniest details to achieve the best bouquet of the drink. But if you use a sensor network, we can "sow" the vineyard wireless sensors and continuously monitor temperature, humidity and other parameters important for the maturation of each vine. A few farms in the valley Villamet (Oregon), Intel researchers have launched a pilot sensor network. The project is led professional ... psychologist Richard Beckwith (Richard Bechwith), research using a technology known as "participant observation". He observes, on the one hand, for the experts who gather information with sensors placed on the plantation, on the other - for the growers themselves, to understand what criteria are important in the field of viticulture for those and others and how This information is linked together [4].

An example of another of the implemented project is the deployment of sensor network based on the Air Force in Florida, USA. The system demonstrated good opportunity to recognize the various metal objects, including moving. Application of sensor network allowed to detect the penetration people and vehicles in a controlled area and track their movements. To solve these problems were used prodigals equipped with magnetoelectric and temperature sensors. Currently, the project scope expanded, and a wireless sensor network is already installed on the range of 10 000x500 m. Appropriate application software developed by several American universities [2].

One of the main problems of the WSN is the limited capacity of batteries installed in motah. Replace the battery on motah, in most cases the problem is rather complicated. That is why to date to reduce energy consumption issues involved large numbers of people. The subjects of works ranges from highly specialized issues associated with the creation of individual components of network facilities (transceivers, microcontrollers, sensors, etc.) with low cost and low power consumption to the problems that arise in the operation of sensor networks (issues relating to the organization of the network, software development, etc.) [5]. But the most important problem is the data. The main problems in this area are:
     - Reducing the cycle of reception and transmission of data;
     - Finding the best routes data;
     - Select the network topology;
     - Resistance to change in the topology;
     - Etc.

Although there are already a large number of developments, but all these developments is not complete, because they can not apply for any number of tasks. For each specific task, it is necessary to organize your approach to resheniyu.Imenno why are the present work is to develop a new protocol will be based on ant algorithms. The main objective for this protocol is the optimal route search data from Motov to the main computer as well as the possibility of choosing the optimal route in the event that one or more Motov failure.

Protocol СТР

The protocol provides data to one of several possible receivers of data, thereby providing a link to one of many at the network level. WSN usually constitutes a tree structure rooted at gateways, and leaves - prodigals. CTP uses frames to update the routing and construction tree network.

At fig.6.2 example tree network. The gateway is the root and other nodes form the set of routes that connect to this root. In the CTP is available addressing, so the node does not send the packet on a particular root, and implicitly selects the root of the tree, by selecting the recipient's next shipment.

Figure 6.2 – Example collection tree.
Figure 6.2 – Example collection tree.

Protocol CTP uses 2 aspect ratio: data frame and frame routing. CTP was designed for networks with relatively low traffic, because Bandwidth channel should be enough to send the frame routing in a time when the transmitted data frames.

CTP uses the expected number of hops (ETX) as the gradient routing. The root has ETX 0. ETX ETX node is a parent plus ETX connection with parent. Ultimately, PAGES, given the current route, select route with a smaller value of ETX. ETX - appears as a 16-bit decimal floating point number of fixed-point within a tenth. For example ETX with a value of 45, appears as ETX 4.5, and the ETX with a value of 10 is presented as ETX 1.

Protocol Tymo

Tymo - an implementation of the protocol Dymo in TinyOS [3], which is a routing protocol for mobile ad-hoc networks. First, the protocol was developed to find routes over IP level. Since the protocol Dymo, a reactive protocol, then he obviously does not keep the network topology. If necessary, send data node establishes a route to the destination. Due to a small exchange of information on routing, reduced network traffic, which saves bandwidth bandwidth. Due to the small volumes of stored information about routing, Dymo easily built in limited mote memory .

When a node is necessary to obtain a route, it distributes Route Request (RREQ) - Package zaporashivayuschy route between the sender and receiver. This package distributed throughout the network or within a certain number of neighboring nodes. When a packet reaches the target (or host has a new route to the goal), the node sends Route Replay (RREP). This response is very similar to Route Request, the only difference being that the package has a unicast route and does not require a response.

When nodes receive a RREQ or RREP, they remain in the cache of information about the sender, so they know the path to the source of the package, which can be used later (if the path is not out of date), without sending RREQ. Nodes are able to accumulate the path followed by the package in the package itself. So manner when the nodes send RREQ or RREP in response package can get information more than the path between two nodes.

When the routes are not used for a long time, they are removed. If the site will ask to send a package through a remote route is generated Route Error (RERR) message to notify the sender node (and other sites) that the requested route is no longer available. As another mechanism service route, Dymo uses serial numbers and counters transitions to determine the usefulness and quality of the route.

Review of systems modeling WSN

Each network requires a different approach to development. That is why the development of sensor networks and software to its great importance to have a system simulation. Modeling systems allow you to develop hardware and software for Motov at much lower cost than in the case use of real devices. So they allow various different parameters, that provides the information necessary to decide advantage of those or other parameters in a given situation. Even more useful fall simulation system for developing and testing protocols data.

At the current time there are a large number of systems modeling as specialized (TOSSIM, SWAN, Em *, GloMoSim, etc.) and general-purpose (NS2, OMNet + +, etc.).

TOSSIM - a simulator applications TinyOS. It is a library of TinyOS. Tossim - simulation of discrete events. Event reports on data from the sensor, on the timer, the packet data or the completion of the calculations. It follows that TOSSIM - emulates the event for wireless sensor systems which prodigals are running TinyOS.

TOSSIM is installed on your computer along with a set of tools needed to create, compile, install and debug applications for wireless sensor networks. Working with data tools via the command line.

The main advantage of TOSSIM is that the emulator allows you to simulate the applications that are written for real Motov. Compiler TOSSIM `and replaces Several low-level application components that interact with the real hardware resources Mota, components that interact with the program implementations of these devices in the emulator.

TOSSIM converts interrupt the computer in the event emulator and arranges them in turn, this turn of events emulator controls the execution of an application for TinyOS separate simulated Mota. Event occurs causing an interrupt handler, which sends signals to the events and calls the team TinyOS, imitating what happens in real motah. These events and the team run TinyOS tasks and initiate the generation of further events emulator.

Another of the well-known simulation environment - is the environment AnyLogic. Wednesday AnyLogic allows the use of different paradigms of simulation (continuous and discrete event simulation, multi-agent approach), and combine them. Important advantages of the system are:
     - Object-oriented approach to modeling;
     - Visual development environment with rich functionality;
     - The possibility of using the language Java.

The system AnyLogic introduced the concept of active object (AO) - extension of the classical objects, which are used in object-oriented languages programming. In AnyLogic active object may also include other elements.

If the standard means AnyLogic not enough to build a model (or their use inconvenient), there is the possibility of using the language Java. In the simplest case, this amounts to a description of actions committed during the transition to another state, triggering a timer or a message arrives. In addition, you can add your own Java code to the active object, and use third-party libraries. This makes the system easily expandable AnyLogic [6].

NS2 - known simulator networks open source. This is a system of discrete event simulation network. NS is widely used for modeling protocols and research networks. The system supports a number of popular network protocols. Allows you to create models of wired and wireless networks. This system allows implementation of virtually any model due to the possibility finish simulator. [9]

Based on the foregoing, we can conclude that at present there are a large number of tools for modeling the FSU. All have their pluses and minuses. For task are the most acceptable simulator TOSSIM. This option is selected for the following reasons:
     - Simulator designed specifically for FSU;
     - Emuliruer that applications written for real Motov;
     - Free distribution;
     - Allows you to emulate the work of the FSU running TinyOS, which is the standard for faketicheski.

Review of methods of interaction sites WSN

To provide for interaction sites FSE can use a sufficiently large number of methods, such as:
     - Probability theory;
     - Computational Mathematics;
     - The theory of stochastic processes;
     - Optimization algorithms, etc.
Consider several existing methods.

The method of collective data. The essence of the method is as follows. Once the network connectivity drops sharply as a result of failure of nodes around the base station, the remaining nodes are combined into clusters and transmit the information gathered collectively, using the principle of coherent information transfer. This significantly increases the distance that can transmit nodes and, thus, can be expected to improve connectivity and extend the lifetime. Under time Life Network is the period of time before the fall of connections below a certain threshold [10].

Algorithm for multi-hop data transmission. The essence of the algorithm lies in the fact that sending messages from one network to another point along the chain of intermediate nodes instead of Direct long-distance transmission. To synchronize the nodes are encouraged to use the recurrent wave propagation service messages, outgoing from the center of the network, in addition to the converging wave data. Oncoming waves must be separated in time to prevent crossings. Optimal is the mode when the control wave is emitted by the base station immediately after the arrival of the wave with the data, because doing so maximizes your width network. The minimum allowable distance between the waves depends on the width of the network and the time of passage through the waves one level. [11]

The method of determining the coordinates in the FSU. The method is to find the geographic location of objects a wireless network, which in turn simplifies formation in the standard ZigBee, routing tables. To implement this method used methods Kolmanovskii filtering and methods of the theory of stochastic processes.

Ant algorithms. These algorithms are widely known. This method gives good results in solving optimization problems of large sizes. Enough worked well for the calculation of telecommunication and computer networks. Currently, good results for solving such complex problems, as the traveling salesman problem, graph coloring, optimization problem of network traffic, etc. [12]

Ant algorithms

The idea of ant algorithm - modeling of the behavior of ants associated with their ability to quickly find the shortest path from the nest to a food source and adapt to changing conditions, finding a new shortest route.
     1. The first ant finds a food source (F) in any way (a), and then returns to the nest (N), leaving behind a trail of pheromones (b).
     2. Then the ants choose one of four possible ways, and then strengthen it and make it attractive.
     3. The ants choose the shortest route, because the longer stronger pheromones have evaporated.
Described above is presented in Fig. 6.3.

Figure 6.3 – Example of finding the shortest route ants.(Animation: volume – 29,5 KB; size –
                       			114x279; number of shots – 7; delay between shots – 1000 ms; delay between first and lst shots – 2000 ms; number of repetition cycles – 6).
Figure 6.3 – Example of finding the shortest route ants.(Animation: volume – 29,5 KB; size – 114x279; number of shots – 7; delay between shots – 1000 ms; delay between first and lst shots – 2000 ms; number of repetition cycles – 6).

Among the experiments on the choice between two paths of unequal length, leading from the colony to a power source, biologists noticed that, as a rule, the ants use shortest route. [6] [10] model of this behavior is as follows:
     * Ant (so-called "Blitz") is random from the colony.
     * If it finds a food source, then returns to the nest, leaving behind a trail of pheromone.
     * These pheromones attract other ants near most likely to go this route.
     * Returning to the nest they will strengthen the pheromone trail.
     * If there are 2 routes, then by a shorter, for the same time, would get more ants than the long.
     * The short route will become more attractive.
     * Long road, eventually disappear due to evaporation of pheromones.

Step in the solution using ant algorithms

In order to construct a suitable ant algorithm for solving any problem, you need:
     1. submit the task as a set of components and conversions, or a set of undirected weighted graphs, where the ants can build solutions;
     2. determine the value of pheromone trail;
     3. heuristics to determine the behavior of ants, when building a solution;
     4. if possible, to implement an effective local search;
     5. select a specific ACO algorithm and apply for solving the problem;
     6. nastroit option ACO algorithm.
     Also, are the determining factor:
     - The number of ants;
     - Balance between the study and use;
     - A combination of greedy heuristics or local search;
     - The moment when the pheromone is updated.

It is expected that further implementation of the algorithm described above will create a new data transfer protocol in wireless sensor networks, which enables find the shortest routes package, with an optimal distribution of information flows across the network.

In the course of research materials were collected and studied matrialy about the technology of wireless sensor networks, methods of data from such systems and of systems for building models of networks. After analyzing all information were identified major problems in the design and use of this technology. On the basis of this was tasked to develop better ways to transfer data.

Were then studied the main system to simulate the work of sensor networks and protocols in them. As a result, had to conclude that, that the most viable options for constructing models are: a system simulation FSE - TOSSIM.

Eventually, after studying one of the most popular algorithms for optimization of various processes, and imeenno ant algorithm, we can conclude that this algorithm will allow us to solve the task.

1. H. Karl and A. Willig. Protocols and Architectures for Wireless Sensor Networks. John Wiley & Sons, May 2005

2. Максим Сергиевский - Беспроводные сенсорные сети
         [Electronic resource] - Access Regime: http://www.compress.ru/Article.aspx?id=17950

3. Description Protocol Tymo
         [Electronic resource] - Access Regime: http://docs.tinyos.net/index.php/Tymo

4. Александр Карабуто - Сенсорные сети
         [Electronic resource] - Access Regime: http://offline.computerra.ru/2004/553/35459/

5. Иванов Евгений Владимирович - Автореферат диссертации на соискание ученой степени кандидата технических наук
         [Electronic resource] - Access Regime: http://aspirantura.mipt.ru/zastchita/avtoreferats/frtk/ivanov_ev

6. Дорошенко А.Е - О моделировании сенсорных сетей средствами высокого уровня
         [Electronic resource] - Access Regime: http://www.gradsoft.ua/rus/whitepapers/sensnet9.pdf

7. Максим Сергиевский - Беспроводные сенсорные сети: эмуляция работы. Часть 4
         [Electronic resource] - Access Regime: http://www.compress.ru/Article.aspx?id=19782

8. Сергей Баскаков и Владимир Оганов - Беспроводные сенсорные сети на базе платформы MeshLogic
         [Electronic resource] - Access Regime: http://www.meshlogic.ru/data/Meshlogic.pdf

9. Official site of NS2
         [Electronic resource] - Access Regime: http://www.isi.edu/nsnam/ns/

10.: К.Г.Мишагин, В.А.Пастухов, А.Н.Садков - Продление времени жизни сенсорной сети с помощью методов коллективной передачи информации
         [Electronic resource] - Access Regime: http://www.wl.unn.ru/lab/Portals/0/Sadkov_t.pdf

11. Гекк М.В., Истомин Т.Е., Файзулхаков Я.Р., Чечендаев А.В. - Адаптивный алгоритм быстрой доставки сообщений по выделенным направлениям для беспроводных сетей датчиков
         [Electronic resource] - Access Regime: http://www.ipmce.ru/about/press/articles/adaptive_algorhytm/

12. Чураков Михаил,Якушев Андрей - Муравьиные алгоритмы
         [Electronic resource] - Access Regime: http://rain.ifmo.ru/cat/data/theory/unsorted/ant-algo-2006/article.pdf

13. Official site of TinyOS
         [Electronic resource] - Access Regime: http://www.tinyos.net/special/mission

While writing the given abstract the master's work has not been completed yet. The final date of the work completed is December, 1st, 2010. The text of master's work and materials on this topic can be received from the author or her research guide after the indicated date.