Tiny OS Component Model Component has Frame storage

Tiny. OS

Component Model • Component has: – Frame (storage) – Tasks (computation) – Command Event Interface Messaging Component Internal Tasks Commands Internal State Events

Tiny. OS Application Component Graph Application Originates Message application Thru-Route Message routing Routing Layer messaging packet byte bit sensing application Messaging Layer Radio Packet Radio byte RFM photo clocks ADC Temp i 2 c SW HW

Related MAC mechanisms

CSMA • Listening to the channel before transmission • Pozitive or negative acknowledgments to signal collusion • Lean toward a fundamental assumption that packet transmissions occur with a stochastic distribution, that is very different the correlated trafic found in sensor networks • Aim to support many point-to-point flows

IEEE 802. 11 • Aims to provide a wireless Ethernet illusion • Design based on assumption of a single cell scenario, with mobile stations always in range of at least one base station – Hand off when migrating from one cell to another – No multihop scenario • Assumes peer-to-peer communications rather than many-to-one data propogation

Bluetooth • Model of creation a “wireless cable” illusion • Primary MAC protocal is a centrialized TDMA protocal within piconet – Relatively static ad-hoc network supporting a small number of nodes within single cell • No multihop scenario • Inappropriate for sensor networks

Applicable Mechanisms • • Listening Mechanism Back off Mechanism Contention Based Mechanism Rate Control Mechanism

Self-Organization of a Wireless Sensor Network

Self-Organization • A self-organized network is an independent collection of nodes in which enough information —or the ability to retrieve such information--is present in order to allow transfer of information between any two nodes in the network. • Either at initialization or after a topologymodifying event • Level can vary depending on the network considered.

Spectrum of Self-Organization

Protocols for Self-Organization of a Wireless Network • Protocols must be able to enable network operation during: 1. start up : nodes are booted up, and network is formed. 2. steady state : energy reservoirs are full, can support all the sensing, signal processing and communication. Multihop network is formed in this mode. 3. failure : re-organization, MAC and routing algorithms for the formation of new links and routes to the sink nodes.

Multihop Network • Can operate in both sensor-to-sink and sinkto-sensor. • Bulk of the traffic will belong to the former. • Significant strain on the energy resources of the nodes near the sink, that neighborhood will be more susceptible to energy depletion and failure.

Energy Conserving Techniques • Sensor nodes will do local processing, as opposed to exchanging raw data over air • Protocols must reduce messaging overhead. • These two will lead to the requirement of highly localized and distributed algorithms for data processing and networking.

Protocols for Self-Organization of a Wireless Network(cont. ) • SMACS(Self-organizing Medium Access Control for Sensor Networks): for network start up and link layer organization • EAR(Eavesdrop-And-Register)Algorithm: enables seamless interconnection of mobile nodes in the field of stationary nodes • SAR(Sequential Assignment Routing): facilitates multihop routing • SWE(Single Winner Election)- MWE(Multi-Winner Election): handle the necessary signalling and data transfer tasks in local cooperative information proccessing.

Link Layer Issues Channel Access Classes: • Contention or explicit organization in time/freq. : not suitable for sensor networks since it requires monitoring channel at all times • Organized channel access: - determines network radio connectivity to discover radio neighbors of each node - assign collision free channels to links * centalized channel assignment * distributed assignment

SMACS= Neighbor Recovery+Channel Assignmet • Infrastructure building protocol that forms a flat topology • A distributed protocol which enables nodes to discover their neighbors and establish transmission/reception schedules for communicating them without the need for any local or global master nodes

EAR(Eavesdrop-And. Register)Algorithm • Offers continuous service to the mobile nodes under both mobile and stationary constraints. • Primary constraint: battery power; mobile and stationary sensors must be established with as few messages transmitted by stationary sensors as possible. • Hand off may not be required. • Mobile nodes have the registry of the neighbors. • Acks are avoided by timeouts, thresholds.

Routing • Multihop Routing - objective: to provide priority service with robustness on a long term basis - more energy will spent on route setup and maintenance • Cooperative Routing - reducing overhead in setup since data traffic is light

Multihop Routing • • Minimum energy per packet Minimum cost per packet Creation of multiple paths Parameters: - energy resources estimated by maximum number of packets - additive Qo. S metric(higher metric= lower Qos) (assumed low mobility)

SAR(Sequential Assignment Routing) • Selection of a path among multipath by the node which generates the packet • Objective: to minimize the average weighted Qo. S metric throuhout the lifetime of the network • Criteria: - energy resource - Qo. S metric - Priority level of a packet

Cooperative Signal Processing • Noncoherent - raw sensor data will be preprocessed to be forwarded to central node - central node selection algorithms: * SWE(Single Winner Election) * ST(Spanning Tree) • Coherent -Limited number of sensor generating data - Explicit computation of minimum energy paths - MWE() is used to decrease energy cost. -Longer delay, higher overhead, lower scalability.

References • Katayoun Sohrabi, Jay Gao, Vishal Ailawadhi, and Gregory J. Pottie, “Protocols for Self organization of a Wireless Sensor network, ” IEEE Pers Commun. , Oct. 2000, pp. 16 -27 • Christopher A. St. Jean, “Self-Organization in Ad Hoc and Multihop Wireless Communication Networks, ” Symposium on Multi-hop/Ad-hoc Wireless Networks, June 2002, France. • J. Jamont and M. Occello, “Using Self-Organization for Functional Integrity Maintenance of Wireless Sensor Networks, ” IEEE Proc. , France, 2003. • R. E. Van Dyck, “Detection Performance in Self-Organized Wireless Sensor Networks, ” National Institute of Standards and Technology Gaithersburg, Maryland, USA

ROUTING • Flooding • Gossiping • Spin • Directed Diffusion • Clustering

Flooding & Gossiping • Flooding: – Diffuse copies of message to all neighbors – Problems: • Implosion • Overlap • Resource Blindness • Gossiping: – Diffuse one copy to random neighbors – Solves implosion problem – Problems: • Overlap

SPIN • Overcome the problems of flooding • Negotiation and Resource-adaptation – Negotiation: helps ensure only useful information will be transferred – Resource manager: keeps track of resource consumption • Disseminate information with low latency and conserve energy at the same time.

SPIN • ADV, REQ, DATA • Spin 1: do not consider energy consumption • Spin 2: if energy is low level, reduce its participation • in terms of energy, – Spin 1 uses 25% as much energy than flooding – Spin 2: 60% meta-data per unit energy than flooding.

Directed Diffusion • Data centric • Attribute-naming • interest including timestamp, gradient, data rate, duration(lifetime) • Reactive routing • Neighbor-to-neighbor • Can be efficient in highly dynamic networks(changes in topology is not important) • trade off some energy efficiency for increased robustness and scale.

LEACH • • • Clustering based Min. Energy dissipation Randomly select nodes as clusterheads Setup & steady phases Clusterhead advertise that they are clusterheads • Based on signal strength(cluster members determined)

References • C. Intanagonwiwat, R. Govindan, D. Estrin and J. Heidemann, “Directed Diffusion for Wireless Sensor Networking”, in IEEE/ACM Transactions on Networking, v. 11, no. 1, February 2003. • J. Kulik, W. Rabiner, and H. Balakrishnan, “Adaptive protocols for information dissemination in wireless sensor networks, ” in Proc. 5 th Annu. ACM/IEEE Int. Conf. Mobile Computing and Networking(Mobi. Com’ 99), Seattle, WA, 1999, pp. 174– 185.

Dynamic Power Management in Wireless Sensor Networks A. Sinha and A. Chandrakasan, IEEE Design Test Comp. , Mar. /Apr. 2001 Massachusetts Institute of Technology

Description • Energy savings via 5 power saving modes • Intermode transition policies investigated

Sensor Network & Node Architecture Nodek Ck R Sensor A/D Micro-OS Strong. ARM Memory Battery and DC/DC converter Radio

Communication Models 1) Direct Transmission 2) Multihop 3) Clustering

Useful Sleep States for the Sensor Nodes Sensor, Sleep State Strong. ARM Memory Radio A-to-D converter s 0 Active On Tx, Rx s 1 Idle Sleep On Rx s 2 Sleep On Rx s 3 Sleep On Off s 4 Sleep Off Tx: Transmit Rx: Receive

Event Generation Model • R: Temporal event behavior over the entire sensing region=> Poisson process with an average event rate lambda-tot • Spatial distribution of events: independent probability distribution PXY(x, y) • pek=prob. that an event is detected by nodek, given the fact that it occurred in R. =………. Pk(t, n)= prob. that “n” events occur in time “t” at node k. Pk(Tth, 0)= prob. of no events occurring in Ck over threshold interval Tth =…………… Pth, k(t)= prob. that at least one event occurs in time t at nodek =1 - Pk(Tth, 0)

State Transition Latency & Power Active ti Idle Active s 0 Pk sk Pk+1 sk+1 t 1 taud, k+1 t 2 tauu, k+1

Steady State Shutdown Algorithm If(event. Occurred()=true){ process. Event(); ++event. Count; lambda_k=event. Count/get. Time. Elapsed(); for (k=4; k>0; k--){ if(compute. Pth(Tth(k)) < pth 0) sleep. State(k); } }

Missed Events ps 4=prob. that no events occur in ts 4, k t s 4 =time duration in s 4 mode =- ln(ps 4)/lamdak Transition Algorithm to almost-off state: No Compute. Pth(Tth(4))<pth 0 Yes No lamdak>0 s 3 Yes Next state test Prob. (1 -ps 4) Sleep? Prob. ps 4 ts 4, k s 4 S 3 Compute