SENSOR NETWORK ARCHITECTURE By Shweta Shrivastava Manali Joglekar Gaurav Rajguru
Outline • Introduction to sensor networks • Evolution • Applications • Architecture • Some commercial sensor networks • The ZigBee Alliance
Introduction • Wireless sensor networks have been identified as one of the most important technologies for the 21st century. • Advances in hardware and wireless network technologies have created low-cost, low-power multi-functional miniature sensor devices. • These networks are revolutionizing sensing in a wide range of applications.
Evolution Of Sensor Networks • Early Research on Military Sensor Networks • During the Cold War, the Sound Surveillance system (SOSUS), a system of acoustic sensors was deployed on the ocean bottom to detect and track Soviet submarines. • National Oceanographic and Atmospheric Administration (NOAA) now uses SOSUS for monitoring oceanic events, e.g., animal activity. • The air defense system with sensors has also evolved over the years.
Evolution Of Sensor Networks ( Contd.) • Distributed Sensor Networks Program at the Defense Advanced Research Projects Agency • Network of many spatially distributed low-cost sensing nodes that collaborate with each other but operate autonomously, with information routed to the nodes that can best use it.
Applications • Infrastructure Security • Environment and Habitat Monitoring • Industrial Sensing • Traffic Surveillance • Military sensing • Parking lot sensor networks
Ad Hoc Sensor Networks • A collection of sensor nodes forming a temporary network in the absence of any stationary infrastructure. • A group of sensors form a clusters. • Each cluster appoints a cluster head to manage its sensors.
Sensor Networks Architecture (contd) • Three layers : • Services layer • Data Layer • Physical Layer • Services include routing protocol, data dissemination and data aggregation. • Data layer models all the messages. • Physical layer consists of nodes that are sinks,children nodes,cluster heads and parents.
Sensor Network Architecture(contd) • The sink nodes broadcast a query. • Sensor nodes close to the sensed object broadcast the sensed data to their neighboring sensor nodes. • Cluster heads receive this data from their children nodes and are responsible for processing and aggregating it . • Cluster heads then broadcast the response to the sink nodes through the neighboring nodes.
Sensor Network Challenges • Low Energy Use • Ad Hoc Network Discovery • Network Control and Routing • Collaborative Information Processing • Tasking and Querying
Low Energy Use • Sensor nodes are deployed in remote areas in many applications. • Servicing of such nodes is a very difficult task. • Hence,lifetime of the node depends on its battery life. This requires very low energy expenditure. • Recharging so many sensor batteries would be expensive as well as time consuming.
Ad Hoc Network Discovery • Network topology frequently changes due to the failure or deployment of new sensors. • Nodes need to know the identity and location of their neighbors for processing and collaboration of information. • Since each sensor node interacts only with its neighbors,global knowledge is not needed. • Hence relative-positioning algorithms can be used.
Network Control And Routing • Network must be able to re-organize itself dynamically in case of node failures or addition of new nodes. • Alternatives to traditional Internet methods e.g., IP is required. • Sensors are deployed in large number • Routes are built using geo-information as and when needed due to data-specific purposes. • IP needs to maintain routing tables and updating it would incur heavy overhead in terms of time, money and energy
Network Control And Routing • Routing algorithms that can adapt to the changes are required. • e.g., Diffusion routing methods that depend only on the information at the neighboring nodes. • A comparison of multicast, flooding and diffusion based routing algorithms was performed and the results showed that multicast protocols require less than half the energy required for flooding and diffusion requires only 60% of the energy needed for even multicast.
Collaborative Information Processing • Nodes collaborate to collect and process data to generate useful information. • Tradeoffs between performance and resource utilization. • Better performance is achieved by processing data from more sensors, but requires more communication resources.(e.g., energy) • Communication of information at low-level(e.g.,raw signals) results in lesser information loss, but requires more bandwidth.
Collaborative Information Processing (contd) • FUSION • Information from multiple sensors has to be fused with the local information. • Information arriving at a node might have traveled multiple paths. • Fusion algorithm should recognize the dependency in the information to be fused and also avoid double counting.
Collaborative Information Processing (contd.) • DATA ASSOCIATION • This problem arises in the presence of multiple targets. • Every node must associate the measured information with individual targets. • Also to avoid duplication and enable fusion, targets detected by different nodes have to associate with each other. • Hence, distributed data association algorithms are required.
Tasking And Quering • Two types of addressing in sensor networks: • Data-centric A query is sent to a specific region. • Address-centric A query is sent to an individual node. • Sensor networks should be data-centric. • Giving unique address to each node is costly. • Limited memory and computational power.
Tasking And Quering(contd) • A sensor field is just like a database in which data is dynamically collected. • The data is distributed across the geographically dispersed nodes. • Hence the need for simple user interface. e.g., handheld unit which accepts speech input from user.
Security • The network should be protected against intrusion and spoofing. • Network techniques need to provide low-latency, survivable and secure networks.
Sensor Networks Communication Architecture. • The sensor nodes are usually scattered in a sensor field as shown in the earlier slide. • Each of these scattered sensors nodes are capable to collect data and route data back to the sink. • Data are routed back to sink by multi hop infrastructureless architecture thru the sink. • The sink may communicate with the task manager node via Internet or Satellite. • The design of sensor networks is influenced by many factors like fault tolerance, scalability, production costs, operating environment, power consumption etc.
The Physical layer • Responsible for frequency selection, carrier frequency generation, signal detection, modulation and data encryption. • So far, the 915 MHz ISM band has been widely suggested for sensor networks. • Energy minimization assumes significant importance in designing the physical layer for sensor networks. • The physical layer is a largely unexplored area in sensor networks. • The open research issues are power-efficient transceiver design and modulation schemes.
The Data Link Layer • The data link layer is responsible for multiplexing of data streams, data frame detection, medium access and error control. • It ensures reliable point-to-point and point-to-multipoint connections in a multi point connections in communication network. • Issues to be considered for the data link layer are: (1) Medium Access Control (2) Error control
Medium Access Control • The MAC protocol in a wireless multihop self-organizing sensor network must achieve the following goals: (1) Creation of network infrastructure (2) Fair and efficient sharing of communication resources between sensor nodes. • Some of the proposed MAC protocols are: (1) Self-Organizing MAC for sensor networks (SMACS) (2) CSMA-Based MAC (3) Hybrid TDMA/FDMA-Based MAC.
Medium Access Control(contd) • The following table shows the qualitative overview of MAC protocols for sensor networks:
Error Control • Another important function of data link layer. • Two important modes of error control are: (1) Forward Error Correction (FEC) (2) Automatic Repeat Request (ARQ) • The usefulness of ARQ in multihop sensor network environment is limited by the additional retransmission, energy cost and overhead. On the other hand, the decoding complexity is greater in FEC since error correction capabilities need to be built in. • So, simple error control codes with low complexity encoding and decoding present the best solutions for sensor networks.
Open Research Issues in Data Link Layer • The key research issues pertaining to the data link layer are: (1) MAC for mobile sensor networks (2) Determination of lower bounds on the energy required for sensor network self-organization. (3) Error control coding schemes (4) Power-saving modes of operation
The Network Layer • The networking layer of the sensor networks is usually designed according to the following principles: (1) Power efficiency is always an important consideration (2) Sensor networks are mostly data centric (3) Data aggregation is useful only when it does not hinder the collaborative effort of the sensor nodes. (4) Attribute based addressing and location based awareness • One important function of the network layer is to provide internetworking with external networks, command and control systems and the Internet.
The Network Layer(contd) • Energy efficient routes can be found based on the available power (PA) in the nodes or the energy required ( ) for transmission in the links along the routes. • Next slide shows the power efficiency of the routes. • An energy efficient route is then selected based on various approaches which are discussed later.
Power Efficiency of Routes. • In the earlier figure, T is the source node that senses the phenomena. It has the following routes to communicate with the sink. (1) Route 1: Sink A-B-T, total PA = 4, total = 3 (2) Route 2: Sink A-B-C-T, total PA = 6, total = 6 (3) Route 3: Sink D-T, total PA = 3, total = 4 (4) Route 4: Sink E-F-T, total PA = 5, total = 6 • We can select an energy efficient route based on various criteria as discussed in the next slides.
Power Efficiency of Routes. • Maximum PA route: The route having maximum PA is preferred. The total PA is calculated by summing PA’s of each node along the route. Accordingly, route 2 is selected. However, we do not select route 2 as it is not power efficient as it only extends route 1 by adding an extra node to it. So we select route 4. • Minimum Energy (ME) route: The route that consumes minimum energy to transmit data packets between the sink and sensor node is ME route. Thus, we see that rout 1 is ME route.
Power Efficiency of Routes. • Minimum hop (MH) route: The route that makes the minimum hops to reach the sink is preferred.. Thus, route 3 is preferred. • Maximum minimum PA node route: The route along which the minimum PA is larger than the minimum PA’s of the other routes is preferred. Thus, route 3 is most efficient.
Current research on Networking Layer. • The various schemes proposed for sensor networks are: (1) Small Minimum Energy Communication Network (SMECN) (2) Flooding (3) Gossiping (4) Sensor Protocols for Information Via Negotiations (SPIN) (5) Sequential Assignment Routing (6) Low Energy Adaptive Clustering Hierarchy (LEACH) (7) Directed Diffusion • The next slide gives an overview of the protocols described above.
Transport Layer • Transport layer is needed when system is planned to be accessed through the Internet or other external networks. • Transport layer protocols are still unexplored. • The protocols may be purely UDP-type protocols, because each sensor node has limited memory and power.
Application Layer • Although many applications of sensor networks are found, the potential application layer protocols for sensor networks remain a largely unexplored region. • It is one of the hottest research issues at present.
Other issues of the protocol stack • The power management plane manages how a sensor node uses its power. • The mobility management plane detects and registers the movement of sensor nodes, so a route back to the user is always maintained and the sensor nodes can always keep a track of who the neighboring sensor nodes are. • The task management plane balances and schedules the sensing tasks given to a specific region.
Conclusions • The flexibility, fault tolerance, high sensing fidelity, low cost and rapid deployment characteristics of sensor networks create many new and exciting areas for remote sensing. • In the future, the wide applications of sensor networks will make it an integral part of our lives.
Some Currently Available Sensors • Smart dust (mote) • Sensor network by Millenial Net • Sensor Network by Ember Corporation
Smart Dust • Developed at UC, Berkeley and now commercialized by Dust Inc., Crossbow etc. • A self-contained sensing and communication platform for a massively distributed sensor network. • Size ~ millimeter scale – around the size of a grain of sand. • Contains sensors, computational ability, bi-directional wireless communications, and a power supply. • Inexpensive enough to deploy by the hundreds. • Have self-organizing capabilities. • TinyOS – An open source OS for motes.
Some Example Applications of Smart Dust • Can be scattered around battlefields to track troop movements. • Can be embedded in roads to collect traffic data. • Used for detecting climatic conditions. • Monitoring energy use in buildings (offices, supermarkets etc.) • Environmental and habitat monitoring (air quality, soil moisture, animal tracking etc.) • Industrial monitoring.