3/13/2002CSE Sensor-Network Schemes1 Sensor-Network Schemes Presented by: Charles ‘Buck’ Krasic Slides adapted from original authors’

3/13/2002CSE Sensor-Network Schemes2 Paper List 1.C. Intanagonwiwa, R. Govindan, D. Estrin, (USC/ISI, UCLA) “Directed Diffusion: A Scalable and Robust Communications Paradigm for Sensor Networks”. MobiCOMM J. Heidemann, F. Silva, C. Intanagonwiwat, R. Govindan, D. Estrin, D. Ganesan, (USC/ISI,UCLA) “Building Efficient Wireless Sensor Networks with Low-Level Naming”. SOSP J. Kulik, W. Heinzelman, H. Balakrishnan, (MIT) “Negotiation-based Protocols for Disseminating Information in Wireless Sensor Networks”. MobiCOMM 1999

3/13/2002CSE Sensor-Network Schemes3 Disaster Response Circulatory Net Embed Embed numerous distributed devices to monitor and interact with physical world: in work- spaces, hospitals, homes, vehicles, and “the environment” (water, soil, air…) Network these devices so that they can coordinate to perform higher- level tasks. Requires robust distributed systems of tens of thousands of devices. The long term goal

3/13/2002CSE Sensor-Network Schemes4 Resource-Adaptive Protocols for Networks of Sensors J. Kulik, W. Heinzelman, H. Balakrishnan, (MIT) “Negotiation-based Protocols for Disseminating Information in Wireless Sensor Networks”. MobiCOMM 1999

3/13/2002CSE Sensor-Network Schemes5 SPIN – Sensor Protocols fro Information via Negotiation J. Kulik, W. Heinzelman, H. Balakrishnan, (MIT) “Negotiation-based Protocols for Disseminating Information in Wireless Sensor Networks”. MobiCOMM 1999

3/13/2002CSE Sensor-Network Schemes6 Overview Motivation and goals Approach to sensor communication: –Meta-data exchanges –Data aggregation –“Resource-Adaptive” applications Implementation using ns Experiments

3/13/2002CSE Sensor-Network Schemes7 Sensor Networks New research area Advantages: –Improved accuracy –Fault tolerance Characteristics: –Wireless network No high-powered central base-station Distribution network –Energy-limited nodes

3/13/2002CSE Sensor-Network Schemes8 System Parameters Quality –Accuracy of result Deadline –Time result required Energy Deadline Quality Goal: Setup framework for analyzing trade-offs

3/13/2002CSE Sensor-Network Schemes9 Classic Network Approaches Flooding –Redundant data transmission Multi-hop routing –Large routing tables –Frequent updates –Complexity Question: Are there better approaches?

3/13/2002CSE Sensor-Network Schemes10 Negotiation Protocol ADV- advertise data REQ- request specific data DATA- requested data AB ADV AB REQ AB DATA Meta-Data Data Naming

3/13/2002CSE Sensor-Network Schemes11 B A Sensor A sends meta-data to neighbor ADV

3/13/2002CSE Sensor-Network Schemes12 Sensor B requests data from Sensor A REQ B A

3/13/2002CSE Sensor-Network Schemes13 Sensor A sends data to Sensor B DATA B A

3/13/2002CSE Sensor-Network Schemes14 Sensor B aggregates data and sends meta- data for A and B to neighbors ADV B A

3/13/2002CSE Sensor-Network Schemes15 All but 1 neighbor request data REQ B A

3/13/2002CSE Sensor-Network Schemes16 Sensor B sends requested data to neighbors DATA B A

3/13/2002CSE Sensor-Network Schemes17 ns Software Architecture RCApplication Resource Manager Network Interface RCAgent Network NeighborEnergy Link Meta-Data Data Meta-Data Data Resource-Adaptive Node

3/13/2002CSE Sensor-Network Schemes18 Resource-Adaptive Application Communication protocol implementation –Internal state –ADV/REQ/DATA algorithm Resource-adaptive decision-making –Application-specific Computation Communication

3/13/2002CSE Sensor-Network Schemes19 Other Simulation Tools Wireless topology generation Radio energy models Statistics collection –Data acquired –Energy dissipated –Redundant data received –Meta-data exchanged

3/13/2002CSE Sensor-Network Schemes20 Test Algorithms Flooding -- Each node floods new data to all of its neighbors. Gossipping -- Each node floods all its data to one, randomly selected neighbor. Negotiating -- nodes decide what data to send based on meta-data advertisements. Sleeping -- Same as negotiating, except that nodes stop sending messages when energy is low. Zzz...

3/13/2002CSE Sensor-Network Schemes21 25-Node Wireless Test Network 70 meters Diameter = 152 meters Node reach = 10 meters Average degree = 4.7 neighbors 59 edges

3/13/2002CSE Sensor-Network Schemes22 Limited Deadline Total Data Acquired Energy Dissipated Time (ms) % Total Data Acquired Total Energy Dissipated (J) Negotiating Flooding Gossipping Sleeping

3/13/2002CSE Sensor-Network Schemes23 Limited Energy Total Data Acquired Time (ms) % Total Data Acquired Flooding Gossipping Negotiating Sleeping Energy Dissipated Time (ms) Total Energy Dissipated (J) 0.5 0

3/13/2002CSE Sensor-Network Schemes24 Data Acquired/Energy Dissipated Flooding Gossipping Negotiating Sleeping Total Energy Dissipated (Joules) % Total Data Acquired

3/13/2002CSE Sensor-Network Schemes25 SPIN Summary Contribution –Sensor networks should be more data-centric (meta-data driven) –Simulation results Advantages: Seems better than flooding Disadvantages: communication still excessive? Future Work: lots!

3/13/2002CSE Sensor-Network Schemes26 Directed Diffusion C. Intanagonwiwa, R. Govindan, D. Estrin, (USC/ISI, UCLA) “Directed Diffusion: A Scalable and Robust Communications Paradigm for Sensor Networks”. MobiCOMM 2000 J. Heidemann, F. Silva, C. Intanagonwiwat, R. Govindan, D. Estrin, D. Ganesan, (USC/ISI,UCLA) “Building Efficient Wireless Sensor Networks with Low-Level Naming”. SOSP 2001

3/13/2002CSE Sensor-Network Schemes27 Directed Diffusion Concepts Application-aware communication primitives –expressed in terms of named data (not in terms of the nodes generating or requesting data) Consumer of data initiates interest in data with certain attributes Nodes diffuse the interest towards producers via a sequence of local interactions

3/13/2002CSE Sensor-Network Schemes28 Directed Diffusion Concepts (cont’d) This process sets up gradients in the network which channel the delivery of data Reinforcement and negative reinforcement used to converge to efficient distribution Intermediate nodes opportunistically fuse interests, aggregate, correlate or cache data

3/13/2002CSE Sensor-Network Schemes29 Illustrating Directed Diffusion Sink Source Setting up gradients Sink Source Sending data Sink Source Recovering from node failure Sink Source Reinforcing stable path

3/13/2002CSE Sensor-Network Schemes30 Local Behavior Choices 1. For propagating interests In our example, flood More sophisticated behaviors possible: e.g. based on cached information, GPS 2. For setting up gradients Highest gradient towards neighbor from whom we first heard interest Others possible: towards neighbor with highest energy 3. For data transmission Different local rules can result in single path delivery, striped multi-path delivery, single source to multiple sinks and so on. 4. For reinforcement reinforce one path, or part thereof, based on observed losses, delay variances etc. other variants: inhibit certain paths because resource levels are low

3/13/2002CSE Sensor-Network Schemes31 Initial simulation studies (Intanago, Estrin, Govindan) Compare diffusion to a)flooding, and b)centrally computed tree (“ideal”) Key metrics: –total energy consumed per packet delivered (indication of network life time) –average pkt delay CENTRALIZED DIFFUSION FLOODING DIFFUSION FLOODING CENTRALIZED

3/13/2002CSE Sensor-Network Schemes32 Experiments on PC104 testbed Initial experimental measurements of diffusion (e.g., for comparison with simulation) –Compare bytes sent by diffusion with and without aggregation (simple in network processing) Measurement Setup –A 5-hop network of 14 nodes on 2 ISI floors (testbed is actually 30 nodes and growing) –Radio: 13kbps radiometrix –1 sink and 1-4 sources (each source sends 112 bytes every 6 seconds)

3/13/2002CSE Sensor-Network Schemes33 Experimental Results Diffusion with suppression Diffusion without suppression Bytes sent by diffusion per event vs. Number of sources

3/13/2002CSE Sensor-Network Schemes34 Comparison to Simulation Diffusion with suppression Diffusion without suppression Bytes sent by diffusion per event vs. Number of sources

3/13/2002CSE Sensor-Network Schemes35 Differences between Simulations and Experiments MAC differences –Modified for simulations to represent hybrid TDMA-Contention –Radiometrix MAC for experiments Channel differences –No obstacles used in ns-2 simulations Note: we have added ability to include simple “terrain” but didn’t try to replicate indoor exp terrain in sims –More packet losses and collisions in experiments Collisions in experiments act as unintentional suppression (make no suppression look better than it will with better mac)

3/13/2002CSE Sensor-Network Schemes36 In network processing: Nested Queries Edge processing overwhelms power and bandwidth consumption Nested queries where low-energy sensors trigger high-energy sensors Edge Processing Nested Queries with In-network Processing

3/13/2002CSE Sensor-Network Schemes37 Experimental Validation: Testbed Measurements Higher delivery ratio for nested query indicates that localizing data traffic benefits performance. % Audio Events Successfully Delivered vs. Number of light sensors 1-level query Nested query

3/13/2002CSE Sensor-Network Schemes38 TinyDiffusion Implementation of Diffusion on resource constrained UCB motes –8bit CPU, 8K program memory, 512 bytes data memory Subset of full system –retains only gradients, and condenses attributes to a single tag. Entire System runs for less than 5.5 KB memory –TinyOS adds ~3.5K and 144 bytes of data. (incl. support for Radio and Photo Sensor) –Diffusion adds ~2K code and 110 bytes of data to TinyOS.

3/13/2002CSE Sensor-Network Schemes39 TinyDiffusion Functionality Resource Constraints –Limited cache size: currently 10 entries of 2bytes each –Limited ability to support multiple traffic streams. Currently supports 5 concurrently active gradients. Tiered Deployment –PC104s running diffusion interface with mote clusters using Tiny D iffusion. –Motes enable dense sensor deployment but can support limited in-network processing –Logical Header format of Tiny Diffusion is compatible with the Diffusion header.

3/13/2002CSE Sensor-Network Schemes40 Gateway Architecture Mote-NIC Serial Device Driver LINUX DIFFUSION Query Data Sink Acoustic Data Source MOTE TINYOS Tiny Diffusion Photo Data Source Data Sink TINYOS Transceiver RFM MOTE ATMEL MHz MCU 8K program memory 512 Bytes Data Memory RFM Radio 900 MHz PC104 AMD Elan™SC400 66MHz CPU 16MB RAM Form Factor: 3.6" x 3.8" x 0.6"

3/13/2002CSE Sensor-Network Schemes41 Tiered Testbed PC-104+(linux) with MoteNIC Tags, Sensor Card UCB Motes w/TinyOS Yet to come: SmartDust (highly specialized nodes) PC/104 Tag UCB Mote

3/13/2002CSE Sensor-Network Schemes42 “Shoebox Testbed v2” Featuring: PC-104+ w/ Pentium 266 Mote-NIC Ethernet for debugging and measurement Linux w/glibc Plastic shoeboxes from local drugstore

3/13/2002CSE Sensor-Network Schemes43 Directed Diffusion: Summary Main contributions –Description of new networking paradigm Interests, gradients, reinforcement –MobiCOMM: simulation results –SOSP: empirical results Advantages –Benefits of in-network processing Aggregation and nested-queries

3/13/2002CSE Sensor-Network Schemes44 Directed Diffusion Summary (cont’d) Disadvantages –Design doesn’t deal with congestion or loss Future Work –Sensor networks today are analogous to the Internet 3 decades ago

3/13/2002CSE Sensor-Network Schemes45 Sensor Card The sensor card is a small (2”x4”) microcontroller board with several on-board sensors and emitters –Microphone –Light sensor –Accelerometer Designed to perform simple sensing tasks at low power. –Currently it is connected to the PC-104 platform by serial. –Data is preprocessed on the sensor board and fed back to the PC-104 for analysis and communication. –The next version of the PC-104 platform will have the capability to be awakened by a peripheral such as the sensor card.

3/13/2002CSE Sensor-Network Schemes46 Reinforced Aggregation Promote In-network Data Aggregation near the Sources for Better Energy Savings Two Approaches for Reinforced Aggregation –Greedy Tree Approach Incremental approach -- Adds minimum number of links on the existing tree –Iterative Approach Selects aggregation points such that energy dissipation for delivering aggregated data is approximately minimized