Sokwoo Rhee, Ph.D. Sheng Liu, Ph.D.

Transcription

1 Sokwoo Rhee, Ph.D. Sheng Liu, Ph.D.

2 A Guide to the Fundamentals of Wireless Sensor Networks by Sokwoo Rhee, Ph.D. and Sheng Liu, Ph.D. Version 1.0 January Millennial Net, Inc. All rights reserved.

3 About the authors Sokwoo Rhee, Ph.D., was a research associate at MIT focusing on wireless biomedical instrumentation. His research led to the development of a sensor ring that measures the wearer s vital signs. The practical application for this ring was nursing home resident care. Residents would wear the rings to monitor their temperature, heart rate, and oxygen saturation; the data would be transmitted continuously to a base station. The small size requirements of the ring necessitated a small battery. While a coin cell battery would fit the bill in terms of size, there were power consumption and mobility challenges still to be met. For practical reasons, the battery would need to run for months not days, the transmission range would need to cover the entire nursing home facility, and the data transmission would need to support mobile residents. Sokwoo began researching mesh networking as the approach to address the requirements and bring the application from a good idea to a practical solution. Sheng Liu, Ph.D., spent five years as a research scientist at MIT, directing industry-sponsored research programs in the areas of controls, robotics, simulations, signal processing, and mechatronics. He then went to Raytheon where, as Senior Development Engineer, he played a critical role in designing receiver spread-spectrum decoding algorithms for differential-gps based aircraft precision approach and landing systems. In 2000, Sokwoo and Sheng joined forces. Sheng s experience in autonomous systems analysis, dynamic programming, and algorithm development combined with Sokwoo s work on the practical biomedical sensor application resulted in the development of a set of innovative techniques to meet the practicality needs of the ring sensor. These techniques broke ground in reducing power consumption, extending transmission distance, and managing a dynamic, mobile network with a high degree of reliability. The result was a wireless sensor networking protocol that was applicable across a wide spectrum of applications. With this breakthrough protocol, Sowkoo and Sheng founded Millennial Net, Inc. and developed and commercialized the Millennial Net wireless sensor networking platform. Today, as chief technology officer and vice president of research respectively, Sokwoo and Sheng continue to lead the industry in developing innovative technologies to enable Millennial Net s customers develop practical wireless sensor networking-enabled applications.

4 1. Introduction...1 Purpose of This Source Book...1 Symbols Used in this Book...2 Defining Wireless Sensor Networks...3 Opportunities...5 Replacing Traditional Wired Networks... 5 New Opportunities Wireless Sensor Networking Overview...6 System Modules...6 Application Platform... 7 Gateway...7 Mesh Node Module... 8 End Node Module... 8 Sensor/Actuator... 8 System Software...9 Module Firmware... 9 API... 9 Network Monitoring System Network Design Considerations...11 Design Drivers Range Shout Versus Whisper Environmental Concerns Radio Frequency Radio Transmission Techniques Power Data Rate Raw Data Rate Network Throughput Duty Cycle Scalability Mobility Mobile Sensors Mobile Gateways Topologies and Data Models...23 Network Topologies Star Mesh Star-Mesh Hybrid Data Models Data Collection Models Bi-Directional Dialogue Data Models Routing Techniques...30 Efficient Protocol Proactive Protocols Reactive Protocols Routing Protocol Design The Millennial Net System...34 Persistent Dynamic Routing Protocol Highly Responsive Reliable Extremely Power Efficient Scalable Build vs. Buy A Complete System System Software Hardware Modules Development and Management Tools Evaluation Kits Glossary...38

5 1. Introduction The Wireless Sensor Networking Source Book provides a methodology for selecting and implementing a wireless sensor network. Your company s project requires integrating a wireless sensor network between a network of sensors and the application used to monitor and control them. You have been put in charge of selecting the wireless sensor network to use. You re familiar with some concepts of wireless networks, but don t feel comfortable enough to make an informed decision on this particular type of wireless system. What do you need to know? What questions need to be asked? Where do you start? You start here with the Wireless Sensor Networking Source Book. This guide is for engineers and decision makers that will be designing, specifying, selecting, or implementing a wireless sensor network. Purpose of This Source Book The source book provides a broad understanding of the technology fundamentals and design considerations that affect function and performance of a wireless sensor network. As opposed to a high-data-rate wireless systems used in LAN applications (WLAN), a low-power wireless sensor network is specifically designed for low-data-rate applications. This guide is designed to provide you with the information necessary to make an informed decision when selecting and integrating such a network system. Table 1-1 provides a quick reference to the information you will find in this guide. 1

6 Table 1-1: Information provided in this guide Guide Section Chapter 2: Wireless Sensor Networking Overview Chapter 3: Network Design Considerations Chapter 4: Topologies and Data Models Chapter 5: Routing Techniques Chapter 6: The Millennial Net System Information Provided This chapter provides a basic understanding of the wireless sensor network building blocks as a prerequisite to a discussion of fundamental network design considerations outlined in Chapter 3. The information presented in this chapter will help you assess the feasibility of a wireless sensor network in your application, to make important scoping and sizing decisions, and to establish a framework to assess different options in specifying and selecting a wireless sensor network system for your application. This chapter provides a look at three textbook topologies and discusses the different data models used by wireless sensor networks to collect and manage data. In this chapter, you ll learn the advantages and disadvantages associated with the different routing techniques developed specifically for wireless sensor networks. This chapter provides a brief overview of Millennial Net s wireless sensor networking platform with Persistent Dynamic Routing. Symbols Used in this Book The symbols shown in table 1-2 are used in this book to illustrate sensor networking concepts: Table 1-2: Symbols used in this guide Symbol Description Sensor/actuator Application End node 2

7 Symbol Description Mesh node Gateway Defining Wireless Sensor Networks Typical wireless sensor network applications share three common requirements: small form factor, long battery life, and dynamic operating environment. Until recently, networks designed for monitoring and controlling sensors or actuators on a network were limited in application and scope due to a major network design consideration the cables required to connect the various sensors and actuators to a centralized collection point. In addition to the costs associated with installing and maintaining communication cables (fiber optic or copper), this type of network infrastructure prevents sensor mobility and severely limits the feasible applications of such a network. Thanks to significant advances in low-power radio and digital circuit design, self-organizing wireless sensor networks are now a reality. Sensors of all types (temperature, motion, occupancy, vibration, etc.) can now be wirelessly enabled and deployed inexpensively and quickly. Wireless sensor networks fundamentally change the economics of deploying and operating a sensor network, unlocking opportunities to achieve new efficiencies in production processes, building control, or monitoring, to name just a few. Wireless sensor networks also enable the development of a brand new 3

8 class of applications and services not previously possible with wired sensor networks. There are no administrative duties associated with establishing and maintaining an ad hoc network. As illustrated in Figure 1-1, wireless sensor networks form what is called a wireless ad hoc network, which refers to a network s ability to self-organize and selfheal. This means there are no administrative duties associated with establishing and maintaining a wireless sensor network. By comparison, a wired infrastructure network, such as the LAN found in most office environments, requires a significant amount of overhead to install and maintain in terms of cabling and administrative time. In an ad hoc network, sensor nodes consisting of a sensor attached to a wireless module can be randomly placed and moved as needed. If the network needs to scale up, additional sensor nodes are easily added. The new sensor nodes and surrounding network will do the work of discovering each other and establishing communication paths through singleand multi-hop paths. All this is made possible through the use of robust, efficient network protocols developed specifically for wireless sensor networks. Figure 1-1: Untethered, mobile ad hoc network nodes 4

9 Opportunities Looking forward, wireless sensor networks will unlock new and exciting applications and services. Today, wireless sensor networks are being used in a number of low-power, low-data-rate applications aiding digital precision instruments on the factory floor, collecting water and gas meter readings, monitoring shipments through the supply chain, and reporting on the vital signs of individual wearers. Looking forward, wireless sensor networks will unlock new and exciting applications and services. Replacing Traditional Wired Networks Sensors and actuators can now be monitored and controlled wirelessly, obviating the expensive installation and maintenance of copper or fiber optic cables. For instance, wireless sensor networks are now being installed in building maintenance systems, replacing the traditional RS-485 cables used to connect the building controller with the various thermostats located throughout a building. The wireless sensor network is transparent to the controller and thermostats, that up until now used the RS-485 cables to communicate with each other. New Opportunities The emerging technology behind wireless sensor networks is opening the door to a new world of opportunities in data collection and system monitoring applications opportunities where traditional wired networks made them economically or physically impossible to consider. Today, applications as varied as monitoring water usage in large apartment complexes to unobtrusively monitoring a patient s blood sugar level are now possible. Wireless sensor networks will allow companies to develop new sources of revenue and cut or eliminate waste. 5

10 2. Wireless Sensor Networking Overview This chapter provides you with a basic understanding of the wireless sensor network building blocks as a prerequisite to a discussion of fundamental network design considerations outlined in Chapter 3. System Modules The modules of a wireless sensor network enable wireless connectivity within the network, connecting an application platform at one end of the network with one or more sensor or actuator devices at the other end. As shown in Figure 2-1, the gateway and end node modules create a transparent, wireless data path between the application platform and sensor. Figure 2-1: Basic wireless sensor network components Exchange of analog or digital information between an application platform and one or more sensor nodes takes place in a wireless fashion. In this example, the data path between the gateway and end node is referred to as a single-hop network link. To extend the range of a network or circumvent an obstacle, a wireless mesh node module can be added between a gateway and an end node as shown in Figure

11 Figure 2-2: Adding a mesh node module This particular example represents a multi-hop data path, in which data packets are handed off from one module to the next before reaching their destination (gateway-to-mesh node-to-end node and vice versa). More elaborate network layouts are discussed later in Network Topologies, but for now, we ll take a closer look at each of the network components shown in Figure 2-2. Application Platform This is the network device (PC, handheld, etc.) used to monitor and control the actions of the various sensors and actuators that are connected to the wireless sensor network. The application platform is capable of making decisions based on the information it gathers from the network. Typically, the wireless sensor network will come with an API (application programming interface) and/or a GUI (graphical user interface) used to interface with the wireless modules. Gateway The gateway is the interface between the application platform and the wireless nodes on the network. The gateway can be a discrete module, or it can be integrated onto a Flash card form factor for a handheld device, for example. All information received from the various network nodes is aggregated by the gateway and forwarded on to the application platform. In the 7

12 reverse direction, when a command is issued by the application program to a network node, the gateway relays the information to the wireless sensor network. The gateway can also perform protocol conversion to enable the wireless network to work with other industry-standard network protocols. Mesh Node Module Hardware design will affect a module s power-efficiency. Considered full-function devices (FFD), mesh node modules (sometimes called routers) are used to extend network coverage area, route around obstacles, and provide back-up routes in case of network congestion or device failure. In some cases, mesh nodes may also be connected via analog and digital interfaces to sensors and actuators, providing the same I/O functionality of an end node module. Mesh nodes can be battery powered or line powered. End Node Module Considered reduced-function devices (RFD), end nodes (sometimes called endpoints) provide the physical interface between the wireless sensor network and the sensor or actuator that it is wired to. End nodes will usually have one or more I/O connections for connecting to and communicating with analog or digital sensor or actuator devices. End nodes are typically battery powered. Sensor/Actuator These are the devices you ultimately wish to monitor and/or control. An example is a sensor monitoring the pressure in an oil pipeline. 8

13 System Software The software required to integrate and operate a wireless sensor network resides as firmware in the system modules and in the application platform as a set of API functions or network monitoring system (NMS). Module Firmware Firmware design will affect a module s power-efficiency. Module firmware is a small, efficient piece of code that incorporates the module into a larger ad hoc network. It drives the module's operation as part of the larger ad hoc network. The firmware is also responsible for packaging the analog and digital sensor data into digital packets and delivering them across the wireless sensor network. API An API, or application programming interface, is a set of commonly used functions for streamlining application development. Used by application developers, an API provides hooks to integrate the application platforms with the modules on the wireless sensor network. API functions are grouped into libraries. In wireless sensor networks, there two different API libraries: High-level Library: These functions are used to integrate the application with the gateway module. Low-level library: These functions are used to integrate the sensor/actuator with the end node module. 9

14 Network Monitoring System A network monitoring system (NMS) is software used to interface with a particular wireless sensor network, eliminating the need for any programming. Through the NMS s graphical user interface (GUI), network operators are able to see the various nodes of their wireless sensor network. Depending on the type of network, control commands can also be issued through the NMS. For example, a pin on a digital interface between an end node and an actuator can be set to high to change the state of the actuator. 10

15 3. Network Design Considerations The information presented in this chapter will help you assess the feasibility of a wireless sensor network in your application, make important scoping and sizing decisions, and establish a framework to assess different options in specifying and selecting a wireless sensor network system. In this chapter, the major design drivers associated with wireless sensor networks are described. Most importantly, you ll learn about the performance trade-offs that may need to be considered during the network design process. Understanding how the design drivers are inter-related will help if and when a trade-off decision needs to be made. Ultimately, this will provide you with the tools needed to design a wireless sensor network that will operate at its optimal level of performance. Design Drivers Table 3-1 contains a matrix used to develop a profile of your particular wireless sensor network application. The profile matrix lists the important network design drivers and will help you determine how important each driver is in the overall design and operation of your network. This exercise will also help determine what design trade-offs may need to be made with each wireless sensor networking system you are investigating. 11

16 Table 3-1: Application profile Design Driver Level of Importance Minimal Moderate Critical Range Short distance between modules Long distance between modules to minimize hops Maximum distance between modules desired Power External power source available Long battery life desired to minimize battery replacement Single battery must provide power for multiple years Data rate Very low data rate Moderate data rate High data rate Duty cycle Low duty cycles Moderate duty cycles High duty cycles Scalability Small network size Moderate network size Large network size Mobility Modules stationary and data paths stable Modules mobile and/or data paths changeable Modules extremely mobile and data paths highly changeable Range The term range can be used to describe either of the following: Network Range: The total physical area covered by a wireless sensor network. Module Range: The distance that data can be transmitted between two modules on a network. Two communicating modules represent the most basic building block in designing a wireless sensor network. Factors that affect the range of a network or network module include: Number of supported network nodes as determined by the manufacturer. 12

17 Power associated with the radio frequency used. Environmental issues, such as walls, electrical interference, etc. By understanding some of the range-related concepts and issues associated with module range, you ll understand how to efficiently attain the desired network range. Shout Versus Whisper Even with the additional modules, the multi-hop whisper method consumes much less power to move data between the two points on a network. When transmitting data between two distant points on a network, more power is not always the best answer to bridging the distance between them. Figure 3-1 illustrates two different methods for transmitting data between two points on a network. Figure 3-1: Shout versus whisper links With the shout method, the modules use high output power to transmit data packets between them. While the two modules are able to communicate effectively, they are not doing it in a very power-effi- 13

18 cient manner. The whisper method illustrates how multiple modules using low output power are used to bridge the same distance. Even with the additional modules, the multi-hop whisper method consumes much less total RF transmit power to move data between the two points on a network. Figure 3-2 illustrates the relationship between power and distance for the two methods. Figure 3-2: Power/distance relationship Multi-hopping is a technique also used in wireless sensor networks to extend the range of a network far beyond the limits of the radio frequency used. If for example, the frequency being used restricted the distance between network modules to no more than feet, this distance could be extended by inserting one or more mesh node modules. The data would then hop from source to destination modules using the mesh nodes as stepping stones. Environmental Concerns The maximum range at which two modules on a network can communicate is affected by a number of environmental and network characteristics. 14

19 Items blocking the line of sight between network modules, such as walls and floors, will limit wireless communications. The type of building material used in such obstacles will affect how well a radio frequency can penetrate the object. Figure 3-3: Penetrating line-of-sight obstacles In cases where obstacles must be circumvented to provide radio connectivity between modules, one or more mesh node modules can be inserted for this purpose as shown in Figure 3-4. Figure 3-4: Circumventing radio-frequency obstacles 15

20 Radio Frequency Each module on the network contains a radio transmitter used to communicate with the other wireless modules on the network. Wireless sensor networks generally use one of the license-free ISM frequency bands. Lower radio frequencies, such as 916 MHz which is license-free in North America, require less power and are better at penetrating objects such as walls or doors. Typically, the radio or RF components consume more than 70% of the total power in full-operation mode, sometimes consuming even more while receiving (RX) than transmitting (TX) data. The RF components also burn significant amounts of power during TX/RX switching or waking up. So, many different scenarios must be considered in the power budget. Radio Transmission Techniques For applications where environmental noise is an issue, the modulation scheme of the radio should also be considered as a way of working around such problems. There are typically two modulation schemes or techniques used for transmitting radio signals over a wireless sensor network one uses narrowband signals while the other transmits wideband or spreadspectrum signals. Narrowband Signals These radio signals use a very narrow portion of the radio frequency bandwidth as shown in Figure 3-4. Spread-Spectrum Signals Spread-spectrum signals are resistant to interference and hard to intercept. A spread-spectrum transmitter takes a narrowband signal and spreads it across a broad portion of the radio frequency in a predefined method. Destination devices receiving the signal understand the pre- 16

21 defined method and de-spread the signal before the data can be interpreted. Spread-spectrum signals are usually created using the direct sequence spread spectrum (DSSS) method or the frequency hopping spread spectrum (FHSS) method shown in Figure 3-4. The DSSS method spreads the narrowband signal out over a broad portion of the frequency band. The FHSS method spreads its signal by hopping the narrowband signal across a broad frequency range as a function of time. Figure 3-4: Narrowband, DSSS, and FHSS signals RF circuitry power consumption is highly dependent on the modulation scheme. Spread-spectrum RF chips consume much more power than typical narrowband radios because of the complex base-band processing. Although spread-spectrum radios offer better immunity to interference, for many sensor network applications, narrowband radios remain a practical and more power-efficient choice. 17

22 Power Power efficiency is a critical design factor for wireless sensor network components. The importance of how efficiently the modules in a wireless sensor network manage their power resources can vary with each application. In some applications, power consumption efficiency is not an issue as access to local power resources is readily available to each module. Modules integrated with the thermostats of a building automation system, for example, can draw their power from the same 24 VAC source used by the thermostats. In other applications, wireless sensor network modules are located in areas where access to local power is not possible, either because of module location or mobility issues. In such instances, the modules typically draw their power from small, coin-cell batteries, making efficient use of power critical. Being able to operate efficiently for long periods of time using battery power is a major advantage of wireless sensor networks over wired networks and a critical design factor. Data Rate Data rate refers to the amount of data the wireless sensor network is capable of carrying or supporting. Expressed in bits per seconds (bps), data rate is evaluated in two ways: the raw data rate and the actual network throughput. Raw Data Rate This value is determined by the radio used in the wireless sensor network modules and is less relevant to wireless sensor network design than network throughput described below. 18

23 Network Throughput This value refers to the actual data rate a network can support. Network throughput is always less than the raw data rate, and will vary based on many factors including: The size of the network (number of nodes). The density of the modules within the wireless sensor network. Modules must negotiate for transmission (TX) time, therefore, the fewer the number of modules within communication range of each other, the faster data can be exchanged. Figure 3-5: Network node density High-Density Network Modules all competing for transmission time. Low-Density Network Little competition between modules for transmission. Duty Cycle The duty cycle of a module refers to the percentage of time the module is active versus inactive, and is determined using the following equation: module time on time period = duty cycle (%) When in an active state, a module is transmitting data, receiving data, or simply listening to the network. Since modules consume power when transmitting or receiving data, it is important to keep the duty cycle to a minimum to achieve the greatest level of power efficiency. When inactive, some end node or mesh node modules can be configured to enter into a sleep mode, conserving valuable energy. These modules will wake up to either issue a network heartbeat on a regular basis or when needed for data transmission and 19

24 reception. The heartbeat is the end node s way of letting the network know it is still there. The duty cycle for the modules on a network should be configurable, especially for powerconscious applications. The duty cycle for the modules on a network should be configurable, especially for power-conscious applications. In some applications, input from a module might only be required every few hours, allowing the module to remain in sleep mode for extended periods of time. The module will wake up briefly to transmit the data it has collected, then return to a power-conserving sleep mode. For applications where module input is required very frequently, keeping the duty cycle low by putting the module in sleep mode even if only for a few seconds enables the modules to conserve a significant amount of energy compared with being constantly in an active or awake state. One key design challenge in reducing duty cycle of mesh nodes modules is to ensure these nodes can wake up in time to route for other mesh nodes. A poorly coordinated sleep/wake-up schedule among mesh nodes can lead to excessive latency or even loss of data. Therefore, reducing duty cycle of mesh nodes must be implemented intelligently in order to save power without sacrificing responsiveness and robustness. Scalability In typical wireless sensor networks, there is an inverse relationship between network size and latency. In typical wireless sensor networks, at any given sampling rate, there is an inverse relationship between network size and latency. In other words, it becomes more difficult to build a responsive ad hoc network as the number of nodes increases. This is 20

25 due to the network overhead that comes with the increased size of the network. In ad hoc networks, the network is formed without any predetermined topology or shape. Therefore, any node wishing to communicate with other nodes should generate more packets than its data packets; these extra packets are generally called control packets or network overhead. As the size of the network grows, more control packets will be needed to find and keep the routing paths. In typical ad hoc networks, the overhead increases exponentially as the network size grows. In a small network, the amount of control packets is almost negligible. But when the network size starts increasing, the overhead increases rapidly. Mobility Mobility refers to the ability of the network to handle mobile nodes and changeable data paths. Mobility refers to the ability of the network to handle mobile nodes and changeable data paths. High network responsiveness is a pre-requisite for supporting mobility. There are two kinds of mobility that a wireless sensor network must support: mobile sensors and mobile gateways. Mobile Sensors The self-configuring nature of ad hoc sensor networks enable them to be able to recognize sensors entering and exiting the network. This enables the network to monitor and control dynamic environments where sensors are not stationary. It also provides low-maintenance scalability. Adding a new sensor to the network requires only placing the sensor node within the network; no further configuration or set up is required. 21

26 Mobile Gateways Gateway mobility enables a gateway device to enter the network, automatically bind to that network and gather data, then leave the network. One mobile gateway can bind to multiple networks and multiple mobile gateways can bind to a given network. 22

27 4. Topologies and Data Models This chapter provides a look at three textbook topologies and discusses the different data models used by wireless sensor networks to collect and manage data. Network Topologies The architectures used to implement wireless sensor network solutions include star, mesh, and star-mesh hybrid topologies. Each of these topologies presents its own set of challenges, advantages, and disadvantages as shown in Table 4-1 and discussed below. Table 4-1: Network topologies Topology Power usage Range Star Low Short Mesh High Long Star-mesh hybrid Low Long Star A star topology, as shown in Figure 4-1, is a singlehop system in which all wireless sensor nodes communicate bi-directionally with a gateway. The gateway can be a PC, PDA, dedicated building control device, embedded Web server, or other gateway to an application platform or another network. The end nodes are identical and the gateway serves both to communicate data and commands among end nodes, and to transfer data to an application or other network, such as the Internet. The end nodes 23

28 do not pass data or commands to each other; they use the gateway as a coordination point. Figure 4-1: Star topology Among wireless sensor networking topologies, the star topology is the lowest in overall power consumption, but is limited by the transmission distance of the radio in each end node back to the gateway. This distance can range from ten to hundreds of meters. Notice also, that there are no alternate communication paths between any of the end nodes and the gateway. Should a path become obstructed, communication with the associated end node may be lost. Mesh Mesh topologies are multi-hopping systems in which all wireless sensor nodes are mesh nodes and communicate directly with each other to hop data to and from the gateway and to pass commands to each other. This is illustrated in Figure 4-2. A mesh network is highly fault tolerant because each sensor node has multiple paths back to the gateway 24

29 or to other nodes. The multi-hop system allows for a much longer range than a star topology. Figure 4-2: Mesh topology Star-Mesh Hybrid A star-mesh hybrid seeks to take advantage of the low power and simplicity of the star topology, as well as the extended range and self-healing nature of a mesh network topology. As shown in Figure 4-3, a star-mesh hybrid organizes end nodes around mesh nodes which, in turn, organize themselves in a mesh network. The mesh nodes serve both to extend the range of the network and to provide fault tolerance. Since end nodes can communicate with multiple mesh nodes, if a mesh node fails or if a radio link experiences interference, the network will reconfigure itself around the remaining mesh nodes. 25

30 Figure 4-3: Star-mesh hybrid topology Data Models The data model is a function of the application and describes the flow of data and how that data is used. The data model characterizes and describes the way in which data flows through and is used in the network, or stated a different way, the interaction between the sensors and the application. Unlike the topology which is a function of the network protocol, the data model is a function of the application. You will need to determine the data model most appropriate for your application based on the application s requirements. Broad categories of data models include data collection and bi-directional dialogue models. 26

31 Data Collection Models Data collection models describe monitoring applications where the data flows primarily from the sensor node to the gateway. Periodic Sampling For applications where certain conditions or processes need to be monitored constantly, such as the temperature in a conditioned space or pressure in a process pipeline, sensor data is acquired from a number of remote sensor nodes and forwarded to the gateway or data collection center on a periodic basis. The sampling period mainly depends on how fast the condition or process varies and what intrinsic characteristics need to be captured. In many cases, the dynamics of the condition or process to be monitored can slow down or speed up from time to time. Therefore, if the sensor node can adapt its sampling rates to the changing dynamics of the condition or process, over-sampling can be minimized and power efficiency of the overall network system can be further improved. Another critical design issue associated with periodic sampling applications is the phase relation among multiple sensor nodes. If two sensor nodes operate with identical or similar sampling rates, collisions between packets from the two nodes is likely to happen repeatedly. It is essential that sensor nodes can detect this repeated collision and introduce a phase shift between the two transmission sequences in order to avoid further collisions resulting in optimal network operation and minimized power usage. 27

32 Event Driven There are many cases that require monitoring one or more crucial variables immediately following a specific event or condition. Common examples include fire alarms, door and window sensors, or instruments that are user activated. To support event-driven operations with adequate power efficiency and speed of response, the sensor node must be designed such that its power consumption is minimal in the absence of any triggering event, and the wake-up time is relatively short when the specific event or condition occurs. Many applications require a combination of event driven data collection and periodic sampling. Store and Forward In many applications, data can be captured and stored or even processed by a sensor node before it is transmitted to the gateway or base station. Instead of immediately transmitting every data unit as it is acquired, aggregating and processing data by remote sensor nodes can potentially improve overall network performance in both power consumption and bandwidth efficiency. One example of a store-and-forward application is cold-chain management where the temperature in a freight container carrying produce or pharmaceuticals, for instance, is captured and stored; when the shipment is received, the temperature readings from the trip are downloaded and viewed to ensure that the temperature and humidity stayed within the desired range. Bi-Directional Dialogue Data Models Bi-directional dialogue data models are characterized by a need for two-way communication between the sensor/actuator nodes and gateway/application. 28

33 Polling Controller-based applications, such as those found in building automation systems, use a polling data model. In this model, there is an initial device discovery process that associates a device ID with each physical device in the network. The controller then polls each device on the network successively, typically by sending a serial query message and waiting for a response to that message. For example, an energy management application would use a polling data model to enable the application controllers to poll thermostats, variable air volume (VAV) sensors, and other devices for temperature and other readings. On-Demand The on-demand data model supports highly mobile nodes in the network where a gateway device enters the network, automatically binds to that network and gathers data, then leaves the network. With this model, one mobile gateway can bind to multiple networks and multiple mobile gateways can bind to a given network. An example of an application using the on-demand data model is a medical monitoring application where patients in a hospital wear sensors to monitor vital signs and doctors access that data via a PDA that is a mobile gateway. A doctor enters a room and the mobile PDA automatically binds with the network associated with that patient and downloads vital sensor data. When the doctor enters a second patient's room, the PDA automatically binds with that network and downloads the second patient's data. 29

34 5. Routing Techniques In this chapter, you ll learn the advantages and disadvantages associated with the different routing techniques developed specifically for wireless sensor networks. A wireless sensor network relies on its network layer's routing algorithm to discover routes and deliver data packets from sources to destinations. The routing layer protocol is also responsible for maintaining and repairing routes when radio links (or hops) along established routes are broken, due to relocation or failure of nodes, sever RF interference, or congestion. It is the routing algorithm that enables a wireless sensor network to self-organize and selfheal. Routing always has some degree of overhead associated with it. The size of this overhead directly affects the responsiveness and scalability of the network. To build and implement highly responsive, efficient, and scalable networks, you need a protocol that is very efficient and minimizes the overhead needed to accomplish its tasks. Efficient Protocol The routing protocol is designed to find the most efficient data path route(s) to use between network modules and to dynamically find new paths when conditions within the network change. The embedded routing protocol used by each of the network modules affects a number of characteristics within the wireless sensor network from the way in which modules self-organize to the way in which data is transmitted from source to destination node. The routing protocol is designed to find the most efficient data path route(s) to use between network modules 30

35 and to dynamically find new paths when conditions within the network change. The more efficient the protocol, the more efficiently the wireless sensor network operates, which results in less power being consumed by each module. A number of highly intelligent routing techniques have been developed over the years for both wired and wireless networks. During the 80s and 90s, substantial advances were made to support the explosive growth of computer networks and particularly, the Internet. Prevalence of wireless communication in recent years further spurs the development of network routing techniques. In general, these routing techniques fall into two main categories: proactive and reactive protocols. Proactive Protocols In proactive protocols, such as the link-state routing and the destination-sequenced distance vector (DSDV) routing, each node in the network maintains route information to every other node, typically in the form of routing tables. These routing tables are updated periodically to account for changes in network topology and link conditions. The main advantage of proactive routing is that route information is constantly updated and, therefore, valid routes are always readily available. The route update process in proactive algorithms, however, requires a significant amount of overhead, consuming network bandwidth. This overhead grows exponentially with network size. Hence, for bandwidth-limited applications such as wireless networks, proactive routing cannot scale. 31

36 Reactive Protocols As opposed to proactive routing, reactive routing algorithms establish and maintain routes on demand (i.e., only at the request of nodes that have traffic to send to specified destination nodes). Reactive routing does not require constant updating of route information, hence, reducing the overhead associated with the update process. The on-demand, dynamic nature of reactive routing makes it particularly effective for ad hoc wireless networks where network nodes can be highly mobile and network connection can be formed in an ad hoc manner without the need of any prescribed infrastructure. Popular routing algorithms, such as dynamic source routing (DSR) and ad hoc ondemand distance vector (AODV) routing, have been applied to an increasing number of ad hoc wireless networking applications such as mobile PC networks (WiFi) and battlefield radio networks. Both DSR and AODV have been adopted as part of the solution framework provided by the Mobile Ad hoc Networks (MANET) task group within the Internet Engineering Task Force (IETF). Routing Protocol Design While the existing routing algorithms are effective for various wired and wireless networking applications, they are not well suited for wireless sensor networks due to their unique characteristics and application requirements that are intrinsically different from the traditional networks. Among others, the main differences between wireless sensor networks and traditional networks are as follows. Nodes in wireless sensor networks run on limited power resources, such as low-capacity batteries. 32

37 Radios used in wireless sensor networks support relatively low data rates, typically in the range of 100K - 200K bits per second. In most cases, wireless sensor nodes operate in license-free radio frequency bands with relatively low output power. As such, radio links among nodes can be easily interfered with and corrupted, especially in popular bands such as the 2.4GHz band used by WiFi, Bluetooth and cordless phones. Wireless sensor nodes typically employ low-cost microcontrollers with limited computation capacity and memory. Many applications require a relatively large-scale network with hundreds of wireless sensor nodes to be densely deployed. Topology of a wireless sensor network may change frequently. These unique characteristics pose significant challenges to routing protocol design for wireless sensor networks. Limited channel data rate and radio output power require a highly efficient routing protocol. With stringent resource constraints, routing in wireless sensor networks must be implemented with minimal duty cycle and overhead. Limited channel data rate and radio output power require a highly efficient routing protocol to carry data with low overhead and direct traffic through reliable, multi-hop routes. In a dynamic network with mobile nodes and strong interference, radio links are constantly broken, and the routing protocol must allow network nodes to quickly repair routes and adapt to changing topology. The routing protocol must also be highly scalable to support formation and maintenance of large-scale networks. 33

38 6. The Millennial Net System The Millennial Net wireless sensor networking system delivers a robust, reliable, scalable networking protocol and a complete system for fast and cost-effective time to deployment. Persistent Dynamic Routing Protocol Millennial Net has developed and optimized its protocol to address the unique characteristics and challenges associated with wireless sensor networking. The end result is a networking system and associated protocol that is highly scalable, ultra-efficient, and extremely responsive and resilient in dynamic environments. The Millennial Net protocol for wireless sensor networks that provides the industry s longest battery life at sensor nodes while delivering data over fault-tolerant links with end-to-end redundancy. The Millennial Net protocol is based on Persistent Dynamic Routing a set of patented techniques for reliable and scalable wireless sensor networks which has been designed specifically to meet all critical challenges of wireless sensor networks. When forming an ad hoc sensor network, Persistent Dynamic Routing requires minimal overhead for requesting and establishing connectivity without relying on the bandwidth-consuming flooding technique. Highly Responsive Self-configuration is initiated from the end nodes (not the gateway) for ultra-efficient, light-weight topology discovery and re-discovery providing high respon- 34

39 siveness for mobile sensors in a dynamic environment. Reliable Persistent routing techniques ensure data packet delivery for highly reliable data transmission required for mission-critical applications. Extremely Power Efficient Dynamic route discovery ensures that the best data route is determined on the fly for efficient bandwidth yielding low power consumption and high battery life. Scalable Low overhead yields high scalability to support a network infrastructure with the headroom to grow and adapt. Build vs. Buy Your ability to take advantage of the benefits of wireless sensor networking is only as good as your ability to quickly and cost-effectively develop and deploy your wirelessly networked sensor application. Many so-called solutions available today are simply chips and stacks, leaving you to source components and fabricate PCBs, optimize networking software, as well as perform integration and application development from scratch. Millennial Net delivers a complete system that lets you concentrate on your application requirements. With the complete system software delivered on hardware modules and APIs to streamline sensor and host application integration, your time to market is fast and your development effort is extremely cost-effective. 35

40 A Complete System Millennial Net delivers a complete system of software, hardware modules, and open system interfaces for fast deployment of robust wireless sensor networks. System Software The Millennial Net system software, based on patented Persistent Dynamic Routing technology forms the foundation of a wireless sensor network that is efficient, responsive, and resilient. Topology discovery is ultra-efficient, light-weight, and highly responsive to mobile sensors and a dynamic environment. Persistent routing techniques ensure reliable data transmission for mission-critical applications. Dynamic route discovery makes the platform extremely scalable and power-efficient providing long battery life. The system is architected to minimize overhead, enabling the network to scale very effectively. Hardware Modules End nodes provide a direct interface to analog and digital sensors. The gateway moves data between end nodes and the host, and monitors data links, devices, and battery status. Mesh nodes extend the range of the network, route around obstacles, and form redundant routes. Development and Management Tools The Sensor Integration API provides sensor-specific functions to streamline the integration process and provide data reduction benefits. The Application Integration API enables quick development of applications to integrate, display, and report on the data collected. The Network Monitoring System is a graph- 36

41 ical application for configuring and monitoring the network. Evaluation Kits Millennial Net offers an Evaluation Kit that lets developers install a wireless sensor network prototype in less than one day. The kit contains everything needed including software (with Millennial Net s patented Persistent Dynamic Routing protocol), hardware, and accessories. More information is available online at 37

42 Glossary API Application Programming Interface: A set of definitions of the ways in which one piece of computer software communicates with another. It is a method of achieving abstraction, usually (but not necessarily) between lower-level and higher-level software. One of the primary purposes of an API is to provide a set of commonlyused functions-for example, to poll a wireless network for active network nodes (mesh nodes and end nodes). Programmers can then take advantage of the API by making use of its functionality, saving them the task of programming everything from scratch. APIs themselves are abstract: software that provides a certain API is often called the implementation of that API. ad hoc network A group of wireless sensors connected as an independent wireless network, communicating directly with each other without the use of an access point. bandwidth The amount of data that can be transmitted in a fixed amount of time. For digital devices, the bandwidth is usually expressed in bits per second (bps) or bytes per second. For analog devices, the bandwidth is expressed in cycles per second, or Hertz (Hz). Bluetooth An industrial specification for wireless personal area networks (PANs). Bluetooth provides a way to connect and exchange information between devices like personal digital assistants (PDAs), mobile phones, laptops, PCs, printers and digital cameras via a secure, low-cost, globally available short range radio frequency. Bluetooth lets these devices talk to each other when they come in range, even if they're not in the same room, as long as they are within 10 meters (32 feet of each other). data model As it pertains to wireless sensor networks, the data model characterizes and describes the way in which data flows through and is used in the network. Common data model categories include data 38