Wireless Sensor Networks Seminar Research Trends in Distributed Systems Florian Schaub Ulm University 19. Nov. 2007 Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 1 / 37
Overview 1 Introduction 2 Characteristics 3 System Software TinyOS 4 Middleware Databases Mobile Agents Events 5 Applications Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 2 / 37
Overview 1 Introduction 2 Characteristics 3 System Software TinyOS 4 Middleware Databases Mobile Agents Events 5 Applications Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 3 / 37
What is Sensing? Measuring real-world phenomena Temperature, humidity, vibration, velocity, light conditions, sound, orientation, weight, gas, chemicals, . . . Many applications • Environmental monitoring (forest vitality) • Weather observation (rain fall in a certain region) • Animal tracking (herd movements) • Warning systems (flood warnings, avalanche warnings) • Industrial sensing (production lines, quality control) • Military applications • . . . Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 4 / 37
Tradtitional Sensing • Autonomous sensor stations • Often expensive and heavy equipment • Wired to infrastructure, or • Manual data collection from stations (e.g. every week) Problems • Cumbersome deployment • Limited coverage • Low sensor density • No real-time data Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 5 / 37
Wireless Sensor Networks • Network of tiny sensor nodes • Battery-powered • Processing capabilites • Cheap • Wireless communication • Self-organizing ad hoc network Advantages • Automatic data collection and reports • Easy deployment in large numbers (e.g. via plane) • Suitable for harsh/hostile environments (e.g. jungle) • Higher sensor density, data accuracy and consistency • Collaborative sensing Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 6 / 37
Problems with Sensor Nodes Restricted ressources • Limited energy (typically 2 AA batteries) • Limited storage capacity • Limited processing capabilites Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 7 / 37
Problems with Sensor Nodes Restricted ressources • Limited energy (typically 2 AA batteries) • Limited storage capacity • Limited processing capabilites Trade-off Autonomous operation for long periods of time vs. limited energy ressources • Strong impact on how sensor nodes operate • Switching off components (incl. radio unit) most of the time • Energy-awareness as the important design factor for WSNs Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 7 / 37
Overview 1 Introduction 2 Characteristics 3 System Software TinyOS 4 Middleware Databases Mobile Agents Events 5 Applications Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 8 / 37
Characteristics of Wireless Sensor Networks Diverse range of application scenarios and requirements But: common tasks for WSNs 1 Sensing real-world phenomena 2 Processing sensor data 3 Communicating with other nodes to share data Resulting in general characteristics for (almost) all WSNs Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 9 / 37
Size and Costs Moore’s Law “Processing power of chips doubles roughly every two years.” Gordon E. Moore, 1965 • Desktop computers and servers • More transistors • Constant physical size • Sensor nodes and embedded systems • Smaller chips • Almost constant computing capabilites • Integration of additional functionality (A/D converters, . . . ) • Minaturization Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 10 / 37
Sensor Nodes dissected Components of a sensor node • Micro-processor chip • One or more microsensors • Radio unit • Data storage • Analog-digital converters • Batteries and antenna Network sensor platforms • Commercially available off-the-shelf products • Cheap production • Low per-unit prices (ca. $150) Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 11 / 37
Network Sensor Platforms Several platforms and vendors • Berkeley Motes Family ( UC Berkeley ), information from www.xbow.com Mote MICA MICA2(DOT) MICAz TelosB Year 2001 2002 2004 2005 CPU ATmega163 ATmega128 ATmega128 TI MSP430 RAM 1 KB 4 KB 4 KB 10 KB Memory 16 KB 128 KB 128 KB 48 KB Flash 32 KB 512 KB 512 KB 1024 KB Bandwidth 40 kbps 38.4 kbps 250 kbps 250 kbps • Other platforms • ESB/ScatterWeb ( FU Berlin ) • MIT Cricket ( MIT ) • Imote 2 ( Intel Research ) Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 12 / 37
Energy Restrictions Energy resources usually restricted (batteries) Lifetime Maximization • Switching off components • Sleep Modes • Communication is most expensive operation • In-node data processing to reduce Communication TelosB energy consumption: Operation Consumption Standby 5 . 1 µ A Idle 54 . 5 µ A Active 1800 µ A Send/Receive 20 , 000 µ A 2.5 years lifetime with 1% sensing, communication every 3 min. Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 13 / 37
Wireless Communication • IEEE802.11 and Bluetooth too energy expensive • Only short range communication (10–15 meters) needed • ZigBee (IEEE802.15.4) more energy-efficient Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 14 / 37
Wireless Communication • IEEE802.11 and Bluetooth too energy expensive • Only short range communication (10–15 meters) needed • ZigBee (IEEE802.15.4) more energy-efficient • Ad hoc communication and self-organization • Neighbor discovery (periodic beaconing) • Negotiation of listening/beaconing intervalls • Additional control communication (e.g. for routing) • Piggybacking control information onto data packets • Gateway nodes • Connected to infrastructure (e.g. satellite, directed radio links) • Nodes route results to gateway nodes Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 14 / 37
Data Aggregation and Dissemination • Simple approach: broadcasting raw data • Network congestion • Nodes have to be sending constantly • Efficient approach: data aggregation • Nodes aggregate received data and own data • Less traffic, less communication • Reduce results to level of interest (compression) • Always process as much data as possible in the network • Utilize collaborative processing power in networks Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 15 / 37
Fault Tolerance and Scalability Fault tolerance • Dynamic network changes (failing nodes, moved nodes, obstacles, . . . ) • Adapt to changes, network reorganization • Data-centric communication (address functionality, not individual nodes) Scalability • WSNs scale from 10 to 1,000 nodes • Efficient routing and communication algorithms • Continous deployment to extend network lifetime • Automatic discovery of new nodes (ad hoc comm., data-centric comm.) Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 16 / 37
Real-time and Security Requirements Real-time Requirements • Time-critical applications rely on real-time data and results • Military applications • Warning systems • QoS assertions (response time, delays) Security • Ensure data authenticity and integrity • Protection against data injection, replay attacks • Tamper resistance of nodes • Protection against eavesdropping or unobtrusive communication • Privacy issues (when nodes are linked to persons) Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 17 / 37
Overview 1 Introduction 2 Characteristics 3 System Software TinyOS 4 Middleware Databases Mobile Agents Events 5 Applications Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 18 / 37
WSN System Software Operating system requirements • High concurrency (sensing, processing, communicating) • Handle data streams (sensor data, communication data) • Energy efficiency • Modularity (keep system small) De facto standard: TinyOS • www.tinyos.net Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 19 / 37
TinyOS Operating system features • Component-based • Event-based programming model • Lightweight multithreading support • System language: nesC extended C with support for network embedded systems TinyOS components • Command handlers (to send commands to lower layers) • Event handlers (to handle lower layer events) • Tasks (specifying program logic) • A fixed-sized memory frame Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 20 / 37
TinyOS System Configuration • Component graph (sometimes called wiring ) • Tiny scheduler • No kernel (!) Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 21 / 37
Overview 1 Introduction 2 Characteristics 3 System Software TinyOS 4 Middleware Databases Mobile Agents Events 5 Applications Florian Schaub (Ulm University) Wireless Sensor Networks 19. Nov. 2007 22 / 37
Recommend
More recommend