energy efficient data dissemination in wireless sensor...
TRANSCRIPT
Energy Efficient Data Dissemination in Wireless Sensor Networks
Dr. Sajid Hussain, Jodrey School of Computer Science,
Acadia University
2Sajid Hussain
OutlinePart 1: Introduction to WSNsPart 2: Coverage, Connectivity, and LocalizationPart 3: MAC, Scheduling, and RoutingPart 4: Latest Interests
3Sajid Hussain
Basic IdeaLarge scale and dense deployment of inexpensive embedded sensors, in order to providein situ, precise, and frequent samples of datafor various applications.
4Sajid Hussain
Wireless Sensor DevicesEmbedded processorMemory/storageRadio Transceiver (10-100 kbps, < 100m)Sensors:
Multi-modal sensing, temperature, light, humidity, pressure, accelerometers, magnetometers, chemical, acoustic, or even low-resolution imagers.
Battery Power Source
5Sajid Hussain
Wireless Sensor Network
Base Station
Sensor nodes
1. Large number of sensor nodes
2. Battery constrained
3. Location aware services
4. Radio communication
5. Persistent queries
Internet
6Sajid Hussain
Ecological habitat monitoringObserver effectUnattended sensors provide cleaner, remote-observer approach to habitat monitoringRich experimental dataAt Great Duck Island, Maine, sensors were deployed in and around burrows of Leach’s storm petrel.
7Sajid Hussain
Structural and Seismic MonitoringMonitoring the conditions of civil structures such as buildings, bridges, roads, and aircraft.Current Techniques: manual and visual inspections or sometimes through expensive and time-consuming technologies, such as X-ray and ultrasound.
8Sajid Hussain
Antarctica - Environmental MonitoringThe wireless sensor network called the 'Unmanned Wireless Intelligent Snow and Ice Observation System in Extreme Environments' has been successfully set up around the Dome-A area near the South Pole, which is the southernmost point on Earth's surface.Dome-A (Dome Argus) is the highest and possibly coldest place in Antarctica and perhaps the coldest naturally occurring place on Earth. Dome-A is the highest ice feature in Antarctica, comprising a dome oreminence of 4.093 m elevation.The system was co-developed by Crossbow Technology and the Chinese Academy of Science (CAS) to develop a solution that can consistently work under extreme environmental conditions such as a 4-month polar-night, -82°C temperature lows and an annual average temperature of -55°C. The wireless system deployed by CAS scientists is designed toovercome the low-temperature, high-altitude and soft snow-surface.
9Sajid Hussain
Antarctica - Deployed SystemThe deployed system consists of two base stations and four nodes. Each node is powered by two low-temperature resistant batteries which are expected to last for one year. The communication capability between each node can achieve ranges of up to 1000m. Each node samplesenvironmental data every 15 minutes, including temperature (weather temperature, snow temperature and the snow temperature below 1 meter), humidity, sunlight, and air pressure. The collected data is sent wirelessly to the central base station that collects and sends the data to Beijing every day. Meanwhile, the other remote base station is used to store the data locally to be collected later by an expedition team. The two base stations ensure that no data is lost in the communication.
10Sajid Hussain
Imagers for Plant Observing SystemsOverview
“Progress was made with new sensor deployments and data collection in three general areas of terrestrial plant ecology at the JamesReserve: modeling the photosynthetic responses of the star moss,Tortula princeps; the photosynthetic responses to sunflecks of bracken fern, Pteridium aquilinum; and using imagers as plant phenology sensors with the pan-tilt-zoom tower cameras.”
Approach“For the mosscam project, we are trying to increase the robustness of modeling photosynthesis using the color changes of the star mossand external sensors for temperature and light. For the bracken fern project, we are measuring the photosynthetic responses of the fronds to light flecks and then will be using a camera to record the light environment in order to estimate light levels the fronds are receiving. For the plant phenology project, we are using daily images of a meadow and target plants in order to eventually use automated processes to detect flowering and leafing events.”
Sourcehttp://research.cens.ucla.edu/projects/2007/Terrestrial/PlantCam/
11Sajid Hussain
Imagers for Plant Observing Systems (cont’d)
12Sajid Hussain
Military Surveillance and Target TrackingCan be rapidly deployed for surveillanceProvide intelligence regarding location, numbers, movement, and identity of troops and vehiclesDetect chemical, biological, and nuclear weaponsDARPA:
US Defense Advanced Research Projects Agency
13Sajid Hussain
Defense and Security (Example)Innovative Wireless Technologies
“Enhanced situational awareness and communications are critical to achieving today's missions at home and abroad. A key aspect of improved situational awareness and communications is sensor networks. These sensors, with ad-hoc networking capability, could provide an early-threat detection sensor network that is rapidly deployable, failsafe and inexpensive. This would of great benefit for both homeland security and military applications.”http://www.iwtwireless.com
14Sajid Hussain
Unattended Ground Services (UGS)I-MeshTM UGS Sensor
Small Size: 3.5” x 5.0” x 7.2”Lightweight: < 5 lbs.Long battery life: up to 2 years, depending on sensor typeFlexible, tri-band radioUp to 50m range between sensorsSpecialized signal processing algorithms for low probability of false alarmsAvailable configured with Acoustic, Infrared, Magnetic or Seismic sensors
I-MeshTM Repeater UnitSmall Size: 3.0” x 5.7” x 8.9”Lightweight, 5 lb. designLong battery life exceeds one yearDual tri-band radios30km range from base station6km range from other repeatersSupports up to 100 sensor nodes
15Sajid Hussain
IndustryRadio Harsh EnvironmentSensors & ActuatorsFault ToleranceIntelligent ControlProcess monitoringImprove the cost and flexibility associated with installing, maintaining, and upgrading wired systems.IEEE 802.15.4 standardZigbee Alliance
16Sajid Hussain
17Sajid Hussain
Design Challenges – Extended LifetimeLimited battery power
Unattended monitoring for several years A typical alkaline battery has 50 watt-hours of energy; less than 1 month of continuous operation.As hardware improvements in battery design are not enough, several protocols are designed for extended lifetime.
18Sajid Hussain
Design Challenges – ResponsivenessReduce duty-cycle for extended network lifetime
Synchronization of sleep schedules is challenging.The long sleep cycles reduce the responsiveness and effectiveness of the sensors.For critical event monitoring, the sleep schedules must be kept within strict bounds, even in the presence of network congestion.
19Sajid Hussain
Design Challenges – RobustnessLarge number of inexpensive and unreliable devices (higher device failure rate).Global performance of the system should not be sensitive to individual device failure; performance should degrade gracefully.
20Sajid Hussain
Design Challenges - SynergyMoore’s law-type advances in technology have ensured that device capabilities in terms of processing power, memory, storage, radio transceiver performance, and even accuracy of sensing improve rapidly (given a fixed cost).Challenge is to design synergistic protocols, which ensure that the system as a whole is more capable than the sum of the capabilities of its individual components.The protocols must provide an efficient collaborative use of storage, computation, and communication resources.
21Sajid Hussain
Design Challenges - ScalabilityExtremely large scale: tens of thousandsProtocols will have to be inherently distributed, involving localized communication, and sensor networks must use hierarchical architectures in order to provide such scalability.Issues: failure handling and in-situreprogramming
22Sajid Hussain
Design Challenges - HeterogeneityDevice capabilities: computation, communication, and sensingSmall number of powerful devices with large number of low-power devices.Two-tier, cluster-based network architecturesSensor fusion techniques because of multiple sensing modalities.
23Sajid Hussain
Design Challenges: Self-configurationUnattended distributed systemsNodes have to be able to configure their own topology; localize, synchronize, and calibrate themselves; coordinate inter-node communication; and determine other important operating parameters.
24Sajid Hussain
Design Challenges - Self-optimizationSignificant uncertainty about operating conditions prior to deployment.In-built mechanisms to autonomously learn from sensor and network measurements collected over time and to use this learning to continually improve performance.The operating environment can change drastically over time.
25Sajid Hussain
Design Challenges - Systematic DesignCan be highly application specificTradeoff between a) ad hoc, narrowly applicable approaches that exploit application-specific characteristics to offer performance gain, and b) more flexible, easy-to-generalize design methodologies that sacrifice some performance.Due to severe resource constraints, systematic design methodologies, reusability, modularity, and run-time adaptation are necessitated by practical considerations.
26Sajid Hussain
Design Challenges - Privacy and SecurityLarge scale, prevalence, and sensitivity of the sensor information require privacy and security.Simple temperature, moisture, and light sensors can result in privacy violations.
27Sajid Hussain
MEMSWhat are MEMS?
MEMS stands for Microelectromechanical systems, a manufacturing technology that enables the development of electromechanical systems using batch fabrication techniques similar to those used in integrated circuit (IC) design. MEMS integrate mechanical elements, sensors, actuators and electronics on a silicon substrate using a process technology called micro-fabrication.
How MEMS work?The sensors gather information by measuring mechanical, thermal,biological, chemical, magnetic and optical signals from the environment. The microelectronic ICs act as the decision-making piece of the system, by processing the information given by the sensors. Finally, the actuators help the system respond by moving, pumping, filtering or somehow controlling the surrounding environment to achieve its purpose.
28Sajid Hussain
Micro/Nano Devices
29Sajid Hussain
Cantilever Response
30Sajid Hussain
Signal Detection
31Sajid Hussain
Chemical Detection
32Sajid Hussain
Chemical Detection & ANNs
33Sajid Hussain
More SensorsEnvironment, Agriculture, Underwater, SeismicHealth-care
Wireless pulse oximeter sensor
Wireless two-lead EKG.Accelerometer, gyroscope, and electromyogram (EMG) sensor for stroke patient monitoring.Courtesy: CodeBlue Harvard
34Sajid Hussain
Radio Frequency Identification (RFID)Passive RFID Tags and RFID readers
Passive tags have no battery so the microchip remains inactive until read with a scanner. The scanner sends a low frequency signal to the microchip within the tag providing the power needed to send its unique code back to the scanner and positively identify the object.Benefits include the reduction of error in recording data, rapid data collection and long term reliability. Inventory management, industry automation, tracking species.
35Sajid Hussain
36Sajid Hussain
Hardware ExamplesMTS310
The MTS310 is a flexible sensor board with a variety of sensing modalities. These modalities include a Dual-Axis Accelerometer, Dual-Axis Magnetometer, Light, Temperature, Acoustic and Sounder. The MTS310 is for use with the IRIS, MICAz and MICA2 Motes.Operating temp. range: -10°C to +60°COperating humidity range 0% RH to 90% RH non-condensing
MTS 400CBDual-axis Accelerometer
Analog Devices ADXL202JE, Acceleration range; resolution: ±2 g; 2 mg at 60 HzNonlinearity: 0.2% of full scale, Zero g bias level: 2.0 mg/°C from 25°C
Barometric Pressure SensorIntersema MS5534AM, Pressure range; resolution: 300-1100 mbar; 0.01 mbarAccuracy: ± 1.5% at 25°C
Ambient Light SensorTAOS TSL2550D, Spectral responsivity: 400-1000 nm, similar to human eye
Relative Humidity & Temperature SensorSensirion SHT11, Humidity range; resolution: 0-100%RH; 0.03% RHAbsolute RH accuracy: ± 3.5% RH, Temp. accuracy: ± 0.5°C @ 25°C
37Sajid Hussain
Hardware Examples (cont’d)MICAz
Processor PerformanceProgram Flash Memory 128K bytesMeasurement Serial Flash 512K bytes, >100,000 measurementsConfiguration EEPROM 4K bytesRAM 4K bytesCurrent Draw: 8 mA Active mode, < 15 μA Sleep mode
RF TransceiverFrequency Band1 2400 MHz to 2483.5 MHz ISM bandTransmit (TX) Data Rate 250 kbpsRF Power 3 dBm (max), 0 dBm (typ)Receive Sensitivity -90 dBm (min), -94 dBm (typ)Adjacent Channel Rejection 45 dB + 5 MHz channel spacing30 dB - 5 MHz channel spacingCurrent Draw:
19.7 mA Receive mode, 11 mA TX, -10 dBm, 14 mA TX, -5 dBm, 17.4 mA TX, 0 dBm1 μA Sleep mode
38Sajid Hussain
Hardware Examples (cont’d)Tmote
250kbps, 2.4GHz, IEEE 802.15.4, Chipcon Wireless Transceiver 8MHz Texas Instruments MSP430 microcontroller (10k RAM, 48k Flash) Integrated onboard antenna with 50m range indoors / 125m range outdoors Integrated Humidity, Temperature, and Light sensors Fast wakeup from sleep (<6us) Hardware link-layer encryption and authentication
39Sajid Hussain
SummaryMEMS/NEMS
Very smallPower: Battery or regularCommunication: wire or wirelessCost: Low, $80-300
RegularLargePower: regular power supplyCommunication: wire or wirelessCost: High, $1000 -10K+
Connectivity, Coverage, and Localization
41Sajid Hussain
Coverage & ConnectivityCoverage
Application-specific quality of information obtained from the environment by the networked sensor devices.
ConnectivityNetwork topology over which information routing can take place.
42Sajid Hussain
Structured vs. Randomized DeploymentRandomized
Futuristic large scale applicationsNodes are dropped from aircraftMixed into concrete
StructuredMedium or small scale applicationsCareful hand placement
1. The location of a sink provides the desired wired network and power connectivity.
2. Sensor nodes are placed in an area where sensor measurements are needed.
3. Add additional nodes to provide requisite network connectivity.
43Sajid Hussain
Network TopologySingle-hop Star
Simple design (++)Poor scalability and robustness (--)
Multi-hop Mesh and GridNecessary for larger areas and networks
Two-tier Hierarchical ClustersAttractive in heterogeneous environmentCluster-head nodes are more powerfulDecomposes larger network into separate zones for local data processing.The second tier nodes can use higher bandwidth or it could be a wired network.
44Sajid Hussain
Connectivity in Geometric Random Graphs
Random Graph ModelA systematic description of some random experiment that can be used to generate graph instances, G=(V,E).A tuning parameter that varies the average density of the constructed random graph.Bernoulli Random Graphs, G(n,p), are formed by n vertices and placing random edges between each pair of vertices independently with probability p.
45Sajid Hussain
Geometric Random Graph G(n,R)G(n,R)
n nodes are placed at random with uniform distribution in a square area of unit size.There is an edge (u,v) between any pair of nodes u and v, if the Euclidean distance between them is less than R.G (n,R) does not show independence between edges. For instance, the probability that edge (u,v) exists is not independent of the probability that edge (u,w) and edge (v,w) exist.
46Sajid Hussain
Connectivity in G(n,R)Network connectivity varies as the radius parameter R is varied. Depending on the number of nodes n, there exist different critical radii beyond which the graph is connected with high probability.Depending on the transmission range, there is some number of nodes beyond which there is a high probability that the network obtained is connected.
47Sajid Hussain
Monotone Properties in G(n,R)A monotonically increasing property is a graph property that continues to hold if additional edges are added to a graph that already has the property.Monotone Property
If the property or its inverse are monotonically increasing.All monotone properties show critical phase transitions; all monotone properties are satisfied with high probability within a critical transmission range.
48Sajid Hussain
Connectivity in G(n,K)G(n,K)
Where n nodes are placed at random in a unit area, and each node connects to its K nearest neighbors.Different nodes can have different powers.K must be higher than 0.074 log n and lower than 2.72 log n.
49Sajid Hussain
Ggrid (n,p,R)In this model, n nodes are placed on a square grid within a unit area, p is the probability that a node is active (not failed), and R is the transmission range of each node.R2 must be Maximum number of hops required to travel from any active node to another is
There exists a range of p values sufficiently small such that the active nodes form a connected topology but do not cover the unit square.
⎟⎟⎠
⎞⎜⎜⎝
⎛
nnO
log
⎟⎠⎞
⎜⎝⎛
nnO log
50Sajid Hussain
Connectivity Using Power ControlExtend the communication range, increasing the number of communicating neighboring nodes and improving connectivity in the form of availability of end-to-end paths.For existing neighbors, it can improve link quality (in the absence of other interfering traffic).It can induce additional interference that reduces capacity and introduces congestion.It can cause an increase in the energy expended.
51Sajid Hussain
Minimum energy connected network construction (MECN)
Minimum Power topologyFor any pair of nodes, there exists a path in the graph that consumes the least energy compared with any other possible path.
EnclosureThe region around a node such that it is always energy-efficient to transmit directly without relaying only for the neighboring nodes within that region.MECN does not necessarily yield a connected topology with the smallest number of edges.
52Sajid Hussain
Minimum common power setting (COMPOW)COMPOW ensures the lowest common power level for maximum network connectivity
First multiple shortest path algorithms (e.g. Bellman-Ford algorithm) are performed, one at each possible power level. Each node then examines the routing tables generated by the algorithm and picks the lowest power level such that the number of reachable nodes is the same as the number of nodes reachable with the maximum power level.Not scalable: each node maintains the state of all nodes.Enforcing common power.
53Sajid Hussain
Minimizing Maximum PowerConnected topology with non-uniform power levels, such that the maximum power level among all nodes in the network is minimized.Suitable for cases where all nodes have the same initial energy level.
54Sajid Hussain
Cone-based Topology control (CBTC)
Each node keeps increasing its transmit power until it has at least one neighboring node in every alpha cone or it reaches its maximum transmission power limit.
55Sajid Hussain
Local minimum spanning tree (LMST)First runs a local minimum spanning tree (LMST) construction for the portion of the graph that is within visible (max power) range.The local graph is modified with suitable weights to ensure uniqueness, so that all nodes in the network effectively construct consistent LMSTssuch that the resultant network topology is connected.
56Sajid Hussain
K-CoverageA field is said to be K-covered if every point in the field is within the overlapping coverage region of at least K sensors.Area: s x sGrid resolution g unit distance(s/g)2 points to examine.Or, enumerate all sub-regions resulting from the intersection of different sensor node-regions and verify if each of these is K-covered, O(n2).
57Sajid Hussain
K-perimeter-coveredDefinition
A sensor is said to be K-perimeter-covered if all points on the perimeter circle of its region are within the perimeters of at least K other sensors.
TheoremThe entire region is K-covered if and only if all n sensors are k-perimeter-covered.
58Sajid Hussain
K-CoveredTheorem
The entire region is K-covered if and only if all intersection points between the perimeters of the n sensors (and between the perimeter of sensors and the region boundary) are covered by at least K sensors.
TheoremIf a convex region A is K-covered by n sensors with sensing range Rs and communication range Rc, their communication graph is a K-connected network graph so long as Rc >= 2Rs.
59Sajid Hussain
Path ObservationGiven a deployment, how well can an adversary with full knowledge of the deployment avoid observation?
Calculate the Voronoi tesselation of the field with respect to the deployed nodes, and treat it as a graph. A Voronoi tesselationseparates the field into separate cells, one for each node, suchthat all points within each cell are closer to that node than toany other node.Label each Voronoi edge with a cost that represents the minimum distance from any node in the field to that edge.Add a starting (ending) node to the graph to represent the left (right) side of the field, and connect it to all vertices corresponding to intersections between Voronoi edges and the left (right) edge of the field. Label these edges with zero cost.Using a dynamic programming algorithm, determine the path between the starting and ending nodes of the graph that maximizes the lowest-cost edge traversed. This is the maximal breach path. The label of the lowest-cost edge is the maximal breach distance.
60Sajid Hussain
Localization
Not always necessaryIn structured, carefully deployed WSN (industrial settings, or scientific experiments), the location of each sensor may be recorded or mapped to a node ID at deployment time.Using satellite-based GPS (expensive) or cellular phone positioning techniques (poor location accuracy – tens of meters).
Reference nodes (static or mobile)Unknown nodes (static or mobile).
Could be cooperative or non-cooperative.
61Sajid Hussain
Key Issues – when to localize?For static environments, may be a one-shot processMobile or dynamic environments, on-the-fly or periodic refreshing
62Sajid Hussain
Key Issues - how well to localize?Resolution of location information desired.Absolute coordinates (x,y,z)Relative coordinates: south of node 24 and east of node 22.Symbolic locations: in room A, in sector 23Absolute locations: as good as +/- 20cm
63Sajid Hussain
Key issues – where to localize?The actual location computation can be performed at several different points in the network:
at a central locationIn a distributed iterative manner within reference nodes in the networkIn a distributed manner within unknown nodes
Location choice factorsResource constraints on various nodesCooperative nodeLocalization technique employedSecurity considerations
64Sajid Hussain
Key issues – how to localize?Different signal measurements can be used as inputs to different localization techniques.
Narrowband radio signal strength readings or packet loss statisticsUWB RF signalsAcoustic/ultrasound signalsInfrared
The signals may be emitted and measured by the reference nodes, by the unknown nodes, or both.The localization algorithm may be used on a number of techniques, such as proximity, calculation of centroids, constraints, ranging, angulation, pattern recognition, multi-dimensional scaling, and potential methods.
65Sajid Hussain
Localization ApproachesCoarse-grained localization using minimal information
A small set of discrete measurementsBinary proximity: can two nodes hear each other?Near-far information: which of two nodes is closer to a given third node?Cardinal direction information: Is node in the north of a given node?
Fine-grained localization using detailed information
Based on measurements, such as RF power, signal waveform, and time stamps.
66Sajid Hussain
Coarse-grained node localization using minimal information
Binary ProximityA set of reference nodes are placed in the environment in some non-overlapping manner.Either the reference nodes periodically emit beacons, or the unknown node transmits a beacon when it needs to be localized.Active Badge Location System
Unique beacon signal once every 15 seconds with a 6 meter range.The active badges, in conjunction with a wired sensor network, provide room level location resolution.
Passive Radio Frequency Identification (RFID) tagsInventory-tracking applicationsUnknown nodes are passive tags and reference nodes belong to a sensor network.
67Sajid Hussain
Centroid CalculationFor high density of reference nodes
Let there are n reference nodes within a proximity of the unknown node (xu,yu)Assumption: each node has a simple circular range R in an infinite square mesh of reference nodes spaced a distance d apart.As R/d is increased from 1 to 4, the average RMS error in localization is reduced from 0.5d to 0.25d.
∑=
=n
iiu x
nx
1
1
∑=
=n
iiu y
ny
1
1
68Sajid Hussain
Geometric Constraints
69Sajid Hussain
Approximate Point in Triangle (APIT)
Location estimates using centroid of an intersection regions.Triangles between different sets of three reference nodes.
Going away from all three vertices of the triangle.
Going away from one but towards the other two vertices of the triangle.
70Sajid Hussain
Radio Signal-based distance-estimation (RSSI)
Pr,dB(d) = Pr,dB (do) – η 10 log(d/do) + Xσ,dB
Pr,dB (d) is the received power at distance dP (do) is the received power at some reference distance doη is the path loss exponentXσ,dB is a log normal random variable with variance σ2 that accounts for fading effects.RF-RSSI offer location accuracy on the order of meters or more.
71Sajid Hussain
Pattern Matching (RADAR)Pre-determined “map” of signal coverage in different locations of the environment.The map is used to determine where a particular node is located by performing pattern matching on its measurements.Limitation: very location specific.
72Sajid Hussain
RF Sequence DecodingUses the relative ordering of received radio signal strengths for different references as the basis for localization.
1. The unknown node broadcasts a localization packet.2. Multiple references record their RSSI reading for this
packet and report it to a common calculation node.3. The multiple RSSI readings are used to determine the
ordered sequence of reference from highest to lowest RSSI.
4. The region is scanned for the location for which the correct ordering of references (as measured in Euclidean distances) has the “best match” to the measured sequence. This is considered the location of the unknown node.
73Sajid Hussain
Network-wide LocalizationSeveral unknown nodes have to be localized in a network with a few reference nodes.Possible scenarios: no reference nodes, a single mobile reference node, or several reference nodes.Constraint-based approaches
MAC, Scheduling, Routing
75Sajid Hussain
Aloha and CSMAUnslotted Aloha
Each node behaves independently and simply transmits a packet whenever it arrives; if a collision occurs, the packet is retransmitted after a random waiting period.
Slotted AlohaAllows transmissions only in specified synchronized slots.
CSMA – Carrier Sense Multiple AccessFirst listen to the channel to access whether it is clear. If the channel is idle, the node proceeds to transmit; otherwise, the node waits a random back-off period and tries again.CSMA is used in IEEE 802.3/Ethernet.
76Sajid Hussain
Hidden and Exposed Node Problems
Hidden node problemPackets collide because sending nodes do not know of another ongoing transmission.
Exposed node problemThere is a wasted opportunity to send a packet because of misleading knowledge of a non-interfering transmission.
It is not the transmitter that needs to sense the carrier, but the receiver. Some communication between transmitter and receiver is needed.
77Sajid Hussain
Hidden and Exposed Node Problems
A B C A B DC
Hidden Node Exposed Node
78Sajid Hussain
Medium Access with Collision Avoidance (MACA)Two control messages to solve hidden and exposed node problems.
RTS - Request to SendCTS – Clear to Send
When a nearby node hears an RTS addressed to another node, it inhibits its own transmission for a while, waiting for a CTS response.If a CTS is not heard, the node can begin its data transmission.If a CTS is received, regardless of whether or not an RTS is heard before, a node inhibits its own transmission for a sufficient time to allow the corresponding data communication to complete.
79Sajid Hussain
MACA - AssumptionsIgnoring the possibility of RTS/CTS collisions.Assuming bidirectional communicationNo packet losses
80Sajid Hussain
IEEE 802.11 MACInfrastructure mode
Single-hop connection to access pointsAd hoc mode
Multi-hop networkDistributed Coordination Function (DCF)
CSMA-CA (collision avoidance) with ACKs.Point Coordination Function (PCF)
A central access point coordinates medium-access by polling the other nodes for data periodically; useful for real-time applications because it can be used to guarantee worst-case delay bounds.
81Sajid Hussain
Distributed Coordination Function (CSMA-CA)A sender first checks to see if it should suppress transmission and back off because the medium is busy; if the medium is not busy; it waits a period DIFS (distributed inter-frame spacing) before transmitting.The receiver of the message sends an ACK upon successful reception after a period SIFS (short inter-frame spacing).The RTS/CTS is used only for unicast packets.Nodes that overhear RTS/CTS messages record the duration of the entire corresponding DATA-ACK exchange in their NAV (network allocation vector) and defer access during this duration.An exponential backoff is used a) when the medium is sensed busy, b) after each retransmission (in case an ACK is not received), and c) after a successful transmission.
82Sajid Hussain
IEEE 802.15.4 MACDesigned for use in low-rate wireless personal area networks (LR-WPAN); most of its unique features are for beacon-enabled mode in a star topology, where it uses a superframestructure.Active Phase
for communication between nodes and the PAN coordinator.Contains 16 slots consisting of three parts: the beacon, a contention access period (CAP), and a collision-free period (CFP) that allows allocation of guaranteed time slots (GTS). The presence of the collision-free period allows for reservation-based scheduled access; nodes can remain sleep and need only wake-up just before their assigned GTS slots.CAP uses CSMA-CA, which allows for a small backoff to reduce idle listening.
Inactive Phase: can be adjusted depending on the sleep duty cycle desired.
83Sajid Hussain
Power Management in 802.11Infrastructure mode
Nodes inform the access point (AP) when they wish to enter sleep mode so that any messages for them can be buffered at the AP.The nodes periodically wake-up to check for these buffered messages.Energy savings are provided at the expense of lower throughput and higher latency.
84Sajid Hussain
PAMAS – Power Aware Medium-Access with Signaling
Extension of MACARTS/CTS is carried out on a separate radio channel from the data exchange.Nodes turn off the radio (go to sleep) whenever they can neither receive nor transmit successfully.Specifically, they go to sleep whenever they overhear a neighbor transmitting to another node, or if they determine through the control channel RTS/CTS signaling that one of their neighbors is receiving.If a transmission is started while a node is in sleep mode, upon wake-up the node sends probe signals to determine: a) the duration of the ongoing transmission, and b) how long it can go back to sleep.A node will only be put to sleep when it is inhibited from transmitting/receiving anyway.
85Sajid Hussain
Minimize the idle reception energy
Asynchronous Use of an additional radio or periodic low-power listening techniques to ensure that the receiver is woken up for an incoming transmission intended for it.
Sleep ScheduleUse periodic duty-cycles sleep schedules for nodes. Most often the schedules are coordinated in such a way that transmitters know in advance when their intended receiver will be awake.
86Sajid Hussain
Asynchronous Sleep TechniquesSecondary Wake-up Radio
Each sensor with two radios.Primary radio is the main radio and is off for most of the times.Secondary radio is a low-power wake-up radio that remains on at all times.
87Sajid Hussain
Low-power listening/preamble samplingReceivers periodically wake-up to sense the channel. If no activity is found, they go back to sleep.If a node wishes to transmit, it sends a preamble signal prior to packet transmission. Upon detecting such a preamble, the receiving node will change to a fully active receive mode.The scheme will also potentially wake-up all possible receivers in a given transmitter’s neighborhood, though mechanisms such as information in the header can be used to put them back to sleep if the communication is not intended for them.
88Sajid Hussain
Low-power listening/preamble sampling
Preamble SamplingActive to Receive Message
Send Data MessagePreambleSender
Receiver
89Sajid Hussain
WiseMACThe long preamble can cause throughput reduction and energy wastage for both sender and receiver.Using additional contents of ACK packets, each node learns the periodic sampling times of its neighboring nodes, and uses this information to send a shorter wake-up preamble at just the right time.Packets also contain a “more” bit, which the transmitter uses to signal to the receiver if it needs to stay awake a little longer in order to receive additional packets intended for it.
90Sajid Hussain
TICER/RICERTICER
The sender sends a sequence of RTS signals followed by a short time when it monitors the channel.When the receiver detects an RTS, it responds right away with a CTS signal.If the sender detects a CTS signal in response to its RTS, it begins transmission of the packet.In other words, the sender sends a sequence of interrupted signals and waits for an explicit signal from the receiver before transmitting.
91Sajid Hussain
TICER/RICERRICER
A source that wishes to transmit wakes up and stays in monitoring state. When it hears a wake-up beacon from a receiver, it begins transmission of the data.The receiver in a monitor state that sees the start of a data packet remains on until the packet reception is completed.
92Sajid Hussain
B-MACLow-power listening (LPL)
Implements the preamble-based wake-up technique
Clear Channel Assessment (CCA)Determines whether the channel is busy or not by examining multiple adjacent samples and using an appropriate outlier detection technique.
ACKsIf enabled, a response is sent immediately after receiving any unicast packet.
93Sajid Hussain
Sleep-Scheduled TechniquesSensor MAC (S-MAC)Timeout MAC (T-MAC)Data-gathering MAC (D-MAC)Delay-efficient sleep scheduling (DESS)Asynchronous Sleep Schedules
94Sajid Hussain
Sensor MAC (S-MAC)Each node sleeps a while, and then wakes up to listen for an interval.The duty cycle is assumed to be same for all nodes.During initialization, nodes remain awake and wait a random period to listen for a message providing the sleep-listen schedule of one of their neighbors.If no message is received, they become synchronizernodes, picking their own schedules and broadcasting them to their neighbors.Nodes that hear a neighbor’s schedule adopt that schedule and are called follower nodes.The boundary nodes may adopt a) multiple schedules, or b) the single schedule of one neighbor.
95Sajid Hussain
Sensor MAC (S-MAC) – cont’dThe nodes periodically transmit these schedules to accommodate any new nodes joining the network.As listening period is on the order of a second, clock drift is not a problem.Adaptive listening: addition of variable length active period
• Similar to 802.11, uses RTS/CTS packets• CSMA• Overhearing avoidance, where interfering nodes are
sent to sleep so long as the NAV (network allocation vector) is non-zero.
• Fragmentation of larger data packets into several smaller ones, for all of which only one RTS/CTS exchange is used.
• Energy savings at the cost of sleep latency!
96Sajid Hussain
Timeout MAC (T-MAC)Similar to S-MACThe length of each cycle is kept constant, but the end of the active period is determined dynamically by the use of a timeout mechanism.If a receiver does not receive any messages (data or control) during the timeout interval, it goes to sleep.If it receives such a message, the timer starts afresh after the reception of the message. This renewal mechanism allows for easy adaptation to spatio-temporal variations in traffic.
97Sajid Hussain
Timeout MAC (T-MAC) – cont’dEarly sleep problem
When a node has to be silent due to contention in a given cycle, it is unable to send any message to its intended receiver to interrupt its timeout.Solution 1:
Use short FRTS (future request to send) control message that can be communicated to the intended recipient asking it to wait for an additional timeout period.
Solution 2: “Full buffer priority”, a node prefers sending to receiving when its buffer is almost full. A node has higher priority to send its own packet instead of receiving another packet, and is able to interrupt the timeout of its intended receiver.
98Sajid Hussain
Delay-efficient Sleep Scheduling (DESS)
Each node picks a unique slot out of k slots to use a reception slot and publishes this to its neighbors.Neighbors use the published slot to transmit to the corresponding node.The path delay is the sum of all the per-hop delays along a given path.The end-to-end delay between the two nodes is the minimum path delay.
99Sajid Hussain
DESS - ProblemGiven a communication graph, assign one of k distinct reception slots to each node in the network so that the maximum end-to-end delay between all pairs of nodes in the network (referred to as the delay diameter of the network) is minimized.Optimal solutions are known in the case of tree- and ring-based graphs, and good approximations can be found for a grid network.Multiple schedules provide further improvements.
100Sajid Hussain
Asynchronous Sleep SchedulesDesign independent sleep-wake schedules for individual nodes that guarantee that the wake-up intervals for neighbors overlap.Potentially requires longer wake-up times, but easy design and implementation.A wake-up schedule function (WSF) fu for a given node u is defined as a binary vector of T slots that indicates the k active slots during which that node u will be awake.The number of overlapping active slots between fuand fv is C(u,v).The necessary condition for C(u,v) >= m is that kv.ku >= m.T. For a symmetric WSF design, therefore, the feasible set satisfies k >= sqrt(m.T)
101Sajid Hussain
Contention-free ProtocolsTDMAFDMACDMANo nodes within two hops of each other may use the same slot; it prevents hidden terminal collisions.
102Sajid Hussain
Stationary MAC and startup (SMACS)Each node need only maintain local synchronization.During the starting phase, each node decides on a common communication slot with a neighboring node through handshaking on a common control channel.Each link also uses a unique randomly chosen frequency or CDMA frequency-hopping code. It is assumed that there are sufficiently many frequencies/codes to ensure that there are no common frequency/time assignments within interference range.The slot is then used periodically, once each cycle, for communication between the two nodes.
103Sajid Hussain
Reservation-based Synchronized MAC (ReSync)Each node maintains the notion of an epoch based on its local time alone, but it is assumed that each node’s epoch lasts the same duration (or can be synchronized with nearby neighbors accordingly).Each node sends a short intent message.By listening for sufficiently long, each node must further learn when its neighbors send intent messages so that it can wake-up in time to listen to them.The intent message indicates the planned transmission time of the data; the intended recipient will then wake-up at the corresponding time.
104Sajid Hussain
Traffic-Adaptive Medium Access (TRAMA)Time epochs are divided into a set of short signaling slots, followed by a set of longer transmission slots. TRAMA consists of three key parts:Neighbor Protocol (NP): Nodes exchange one-hop neighboring information during the random access signaling slots. The signaling slots are sufficiently long to ensure that this information is reliably transmitted and all nodes have consistent two-hop neighborhood information.Schedule Exchange Protocol (SEP): Each node publishes it schedule during its last winning slot in each epoch. The changeover slot is used for synchronization: therefore all neighbors are required to wake-up to listen to it. The node sleeps at all times when it is not required to transmit or receive.Adaptive Election Algorithm (AEA): Uses a hash function based on node IDs and time to ensure that there is a unique ordering of node priorities among all nodes within any two-hop region at each time.
105Sajid Hussain
Topologies for ConnectivityBasic/adaptive fidelity energy conserving algorithms (BECA/AFECA)Geographic Adaptive Fidelity (GAF)Adaptive Self-Configuring Sensor Network Topology Control (ASCENT)Neighborhood Coordinators (SPAN)
106Sajid Hussain
BECA – Basic Energy Conserving Algorithms
Active
ListeningSleep
Send local dataAfter Ta units with no traffic
Send local data
After Tl
After Ts
AFECA: sleep time is chosen as the basic sleep time multiplied by arandom number between 1 and N, where N is the number of neighborsestimated in the listening period.
107Sajid Hussain
GAF – Geographic Adaptive Fidelity
Active
DiscoverySleep
After Td
Receive discovery msg from high rank nodes.
After Ta
After Ts
Goal: A single node with the highest remaining lifetime is awake in each virtual grid.
All nodes in one grid can communicate with all others in the adjacent grid.
Nodes in discovery/active state transmit discovery messages containing nodeID, gridID, and estimated residual lifetime.
108Sajid Hussain
CEC – Cluster-based Energy ConservationA set of clusters is created such that each has an elected cluster-head.A cluster-head is a node with the greatest residual lifetime among its neighbors.To ensure connectivity, additional gateway nodes are elected – primary gateway nodes directly connect two distinct cluster-heads, while secondary gateway nodes connect to other cluster-heads through other secondary/primary gateway nodes.All nodes that are not cluster-heads or gateway nodes are eligible to sleep.The sleep timers are set so that nodes wake-up to run a re-election before the cluster-head’s energy is depleted. Nodes that are asleep may wake-up to send their own information, and any data intended for them must be buffered for pick-up after they are awake.
109Sajid Hussain
ASCENT
Sleep
TestPassive
After Tp
Neighbors < NT and (loss rate > LT or help)
After Tt
Neighbors > NT or leads to increased loss rate
For highly dynamic environments; nodes wakeup to assist in routing depending on the number of active neighbors and the measured data loss rates in their vicinity.
Help messages are sent by a node that needs the assistance of neighboring nodes to help relay messages to a given destination.
Active
After Ts
110Sajid Hussain
SPANActive/Coordinator
TestSleep
Every pair of neighbors can reach each other directly or via some other coordinators.
Two neighbors cannot reach each other directly or via one or two coordinators.
Periodically
The decision regarding eligibility rule is made based on the content of HELLO packets sent by all nodes in which they announce their current coordinators and current neighbors.
A randomized prioritized backoff is used to ensure that multiple nodes do not elect to become coordinators simultaneously.
111Sajid Hussain
PEAS – Probing Environment and Adaptive Sleep
Active
ProbeSleep
After randomized sleeping time
No REPLY for PROBE
Receive REPLY
The node uses a randomized timer with exponential distribution of rate λ to transition to the probe state.
In the probing state, a node detects if there is any active node within a probing range Rp, by sending a PROBE message at the appropriate transmit power and listening for REPLY messages.
If there are no responses for PROBE messages, the node transitions to the active state and stays there until its energy is depleted.
112Sajid Hussain
Sponsored Sector
Ready-to-off
Ready-on-duty
Off-duty / Sleep
IneligibleAfter Tw without receiving SAM
Next round
Each node checks to see if its coverage area is already covered by active neighbors before deciding whether to be on or off.
SAM: status advertisement message
On-duty / Active
Receiving SAM/ineligible
Eligible for Td / broadcast SAM
113Sajid Hussain
Objectives - Energy Efficient CommunicationDelay the first node deathDelay the death of last 10% of nodesUniform distribution of loadMaintain given coverage while there is decrease in connectivityUse correlation of data to avoid unnecessary transmissionsPowered nodes (transmitters/repeaters) for energy efficiencyTwo-tier architectureSensing range is the limiting factor as compared to transmission range.
114Sajid Hussain
Routing Matrix
A B
C DE
F
dAC=0.9
dCD=0.9 dDE=0.9
dEB=0.9
dAB=0.1
dAF=0.8dFB=0.8
Obvious choice: Direct from A to B, 10 retransmissionsLong path: through C, D, and EThird path: through F
115Sajid Hussain
Metrics for energy-reliability tradeoffs
For rapid link quality fluctuations:environment containing highly mobile objectsNodes are mobileC = log (1 + |f|2 / dn . SNR)C: channel capacityf: fading state of the channeld : distance between transmitter and receiverSNR: signal-to-noise ratio without fadingn: path loss component
116Sajid Hussain
Relay Diversity
A
D
E
B
C
117Sajid Hussain
Extreme Opportunistic Routing (ExOR)
The node closest to the destination that receives a given packet will forward the packet further.Priority Ordering:
At each step, the transmitter includes in the packet a schedule describing the candidate set of receivers and the priority order in which they should forward the packet.
118Sajid Hussain
Extreme Opportunistic Routing (ExOR)Transmission acknowledgements:
A distributed slotted MAC scheme is used whereby each candidate receiver sends the ID of the highest-priority successful recipient known to it. All nodes listen to all ACK slots. This constitutes a distributed election procedure whereby all nodes can determine which node has the highest priority among the candidates that received the packet successfully. Even if a node does not directly hear the highest-priority node’s ACK, it is likely to hear that node’s ID during another candidate’s ACK.
119Sajid Hussain
Extreme Opportunistic Routing (ExOR)
Forwarding Decision:After listening to all ACKs, the nodes that have not heard any IDs with priorities greater than their own will transmit. There is a small possibility that ACK losses can cause duplicate transmissions; these are further minimized to some extent by ensuring that no node will ever retransmit the same packet twice (upon receiving it from different transmitters).
120Sajid Hussain
Gradient Cost Routing (GRAd)All nodes maintain an estimated cost to each active destination (this set would be restricted to the sinks).When a packet is transmitted, it includes a field that indicates the cost it has accrued to date (i.e. number of hops traversed), and a remaining value field, that acts as a TTL (time-to-live) field for the packet. Any receiver that receives this packet and notes that its own cost is smaller than the remaining value of the packet can forward the message, so long as it is not a duplicate.
121Sajid Hussain
Gradient Broadcast Routing (GRAB)GRAB also maintains a cost field through all nodes.The packet travels from a source to the sink, with a credit value that is decremented at each step depending on the hop cost.An implicit credit-sharing mechanism ensures that earlier hops receive a larger share of the total credit in a packet, while the later hops receive a smaller share of the credit.An intermediate node with greater credit can consume a larger budget and send the packet to a larger set of forwarding eligible neighbors.
122Sajid Hussain
Power-aware RoutingIf all links require the same energy for transmission, the minimum hop-count routing could result in minimum energy expended per packet.If different links have uneven transmission costs, the route that minimizes the energy expended in end-to-end delivery of a packet would be the shortest path route computed using the metric Ti,j.However, in networks with heterogeneous energy levels, this may not be the best strategy.
123Sajid Hussain
Lifetime-maximizing RoutingA technique is needed that balances the two goals: selecting the minimum energy path when all nodes have high energy at the beginning, and avoiding the low residual energy nodes towards the end.Energy metric is a function of: the transmission cost, the residual energy of the transmitting node, and the initial energy of the transmitting node.
124Sajid Hussain
Load-balanced Energy-aware RoutingNodes forward packets only to neighbors that are closer to the destination.For each neighbor x in its set of candidate neighbors, node y assigns a forwarding probability that is inversely proportional to the cost to destination.Each time the node needs to route any packet, it forwards to any of its neighbors randomly with the corresponding probability. This provides for load balancing, preventing a single path from rapidly draining energy.
125Sajid Hussain
Flow Optimization FormulationIn each iteration, every origin node computes the shortest cost path to its destination, and augments the flow on this path by a small step.After each step, the costs are recalculated, and the process is repeated until any node runs out of its initial energy.The linear program maximizes the total data gathered during the time duration T. It incorporates the following constraints:
A flow conservation constraintA per-node energy constraintA shared bandwidth constraint
126Sajid Hussain
Sajid Hussain and Obidul Islam, An Energy Efficient Spanning Tree Based Multi-hop Routing in Wireless Sensor Networks, Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC), IEEE Communication Society, Hong Kong, China, March 11-15, 2007
Energy Efficient Spanning tree based Routing (EESR)
Follows a greedy approachGenerates balanced and energy efficient aggregation treesConsiders residual energy of both transmitting and receiving nodes
127Sajid Hussain
Edge weight assignmentwij (k) = min{Ei − Tij (k) ,Ej − Rj (k)}Consider both sender and receiver node residual energy Builds asymmetric edge weight Select max(wij) to forward dataAvoids receiving nodes to become overloaded
Energy Efficient Spanning tree based Routing (EESR)
128Sajid Hussain
EESR tree Example
0.3x0.250.23
0.30.25x0.22
0.350.20.2x1
BS321
Receiving Nodes
Send
ing
Nod
es
1 2
3 BS
E=2.3, R=1, C=nil E=2, R=2, C=nil
E=2.1, R=3, C=nil
(b) Example network(a) Transmission cost to send fixed packet length
1 2
3 BS
E=2.3, R=1, C=nil E=2, R=2, C=nil
E=2.1, R=3, C=nil
1.8
1.7
1.75
(c) Node 2 is the weakest; edge weights of node 2 are computed
1 2
3 BS
E=2.2, R=2, C={2} E=1.8, R=2, C=nil
E=2.1, R=3, C=nil
1.551.9
1.8
(d) Node 1 is selected as parentof node 2, and edge weights arecomputed for node 3.
1 2
3 BS
E=2.1, R=2, C={2, 3} E=1.8, R=2, C=nil
E=1.9, R=2, C=nil
1.7
1.751.8
(e) Node 1 is selected as parent of node 3, and edge weights are computed for node 1.
1 2
3 BS
E=1.75, R=2, C={2, 3} E=1.8, R=2, C=nil
E=1.9, R=2, C=nil
(f) BS is selected as parent of node 1, in order to avoid loop.
129Sajid Hussain
Genetic algorithm Genetic algorithm (GA)
GA uses the concept of evolution (survival of the fittest)The key objective is to simulate the process in natural systemsUseful when search space is huge Involves large number of variables
Main components:Chromosome PopulationFitnessSelection Crossover
130Sajid Hussain
GA based Aggregation TreeObjective: generate aggregation trees to maximize lifetimeChromosome: node based encoding
Gene index: represents a node idGene value: parent of that node
Single point crossover; mutationRepair function: eliminates cycles
Obidul Islam and Sajid Hussain, Genetic Algorithm for Data Aggregation Trees in Wireless Sensor Networks, Third International Conference on Intelligent Environments, Ulm, Germany, September 24-25, 2007.
Sajid Hussain and Obidul Islam, Genetic algorithm for energy efficient trees in wireless sensor networks, Advanced Intelligent Environments, Springer, 2008, 13 pages (press).
131Sajid Hussain
Chromosome and Crossover
20 18 05 16 ... 17 250 1 2 3 30 31
Chromosome
1
2
0 3
4
Parent 1
1
2
0 3
4
Parent 2
1
2
0 3
4
Child 1
1
2
0 3
4
Child 2
1
2
0 3
4
Invalid chromosome
1
2
0 3
4
After repair
1 3 1 4 --0 1 2 3 4
1 3 4 2 --0 1 2 3 4
Parent 1
Parent 2
1 3 1 2 --0 1 2 3 4
1 3 4 4 --0 1 2 3 4
Child 1
Child 2
132Sajid Hussain
Fitness
Fitness parameters:Ratio of energy to load (L):Edge weight (E_weight)Residual energy of the weakest node (Rmin): (ei – loadi)
Fitness function:fi = w1·L + w2·E_weight + w3·Rmin
Wcurrent =
∑i
i
loade
previousfpreviouscurrentprevious
e
ffw−+
−+
1
133Sajid Hussain
Impact of population size on fitness value
0 50 100 1500
200
400
600
800
1000
1200
1400
1600
1800
2000
Population size
Fitn
ess
valu
e
100X100m2 network field with 100 sensors
134Sajid Hussain
Impact of number of generations on fitness value
100X100m2 network field with 100 sensors
0 50 100 150 200 250 300 3500
500
1000
1500
2000
2500
Number of generations
fitne
ss v
alue
135Sajid Hussain
Simulation environmentJava based custom simulator
Verification of algorithm
J-Sim : Java based wireless network simulatorMAC effectComponents are written in Java and glued together using Jacl (Tcl/Java)
Performance varies in the MAC based simulator
Obidul Islam and Sajid Hussain, Effect of Layers on Simulation of Wireless Sensor Networks, in Proceedings of the Third International Conference on Wireless and Mobile Communications (ICWMC), IEEE Computer Society, Guadeloupe, March 4-9, 2007
136Sajid Hussain
Custom simulator: offline schedulingJ-Sim simulator: online scheduling
Schedule generator Wireless Sensor Network
(T,f)
Information about current energy level
(EESR, PEDAPPA, GA) J-sim environment
Online scheduling
Simulation environment
137Sajid Hussain
Impact of base station’s location (EESR)
Energy level after the first node death, 100m×100m field, BS at (200, 200)
0
20
40
60
80
100
0
20
40
60
80
1000
0.05
0.1
X axisY axis
Z axis (Joule)
BS direction
138Sajid Hussain
Impact of base station’s location (GA)
Energy level after the first node death, 100m×100m field, BS at (200, 200)
0
20
40
60
80
100
0
20
40
60
80
1000
0.1
0.2
0.3
0.4
Ene
rgy(
J)
X axisY axis
BS Direction
139Sajid Hussain
Impact of base station’s location (PEDAPPA)
Energy level after the first node death, 100m×100m field, BS at (200, 200)
0
20
40
60
80
100
0
20
40
60
80
1000
0.2
0.4
Ene
rgy(
J)
X axisY axis
140Sajid Hussain
Standard deviation of residual energy: SDR
50 60 70 80 90 1000
0.005
0.01
0.015
0.02
0.025
0.03
Number of nodes
stan
dard
dev
iatio
n(J)
EESRGAPEDAPPA
50 60 70 80 90 1000
0.05
0.1
0.15
0.2
0.25
Number of nodesst
anda
rd d
evia
tion(
J)
EESRGAPEDAPPA
(a) 100m×100m field, BS at (50, 50) (b) 100m×100m field, BS at (200, 200)
Standard deviation of residual energy (SDR), 100m×100m field
141Sajid Hussain
RSSI Variation
142Sajid Hussain
RSSI EstimationReceived Signal Strength Indication (RSSI) is used to determine which power level setting is needed to transmit directly between two nodes. The model chosen for our implementation is the Radio Irregularity Model (RIM) because of its ability to simulate differences in sending power amongst different pieces of hardware and anisotropic path loss.
RSSI = Sending Power – Path Loss + Fading
143Sajid Hussain
Degree of Irregularity (DOI)
G. Zhou, T. He, S. Krishnamurthy, and J. A. Stankovic. Impact of radio irregularity on wireless sensor networks. In MobiSys’04: Proceedings of the 2nd international conference on Mobile systems, applications, and services, pages 125–138, New York, NY, USA, 2004.
144Sajid Hussain
Transmission CostPT = PTO + PPA
Transmission Cost, PT
Transmission Startup (opening) cost, PTO
Transmission Amplifier Cost, PPA
PTO = V x IROON x TStartupCurrent Voltage Level, LCurrent Usage (radio ON, oscillator ON), IROON
Time needed to start the radio, Tstartup
PPA = V x Iplevel x (L/R)Packet Length, L (bits)Transmission Rate, R (bits/second)Iplevel (mA)
145Sajid Hussain
Reception CostThe reception cost is calculated using data sheet specifications.For CC2420 specifications, the nominal current is 19.7 mA for reception; whereas 17.4 mA is required for transmitting at the highest power level.
146Sajid Hussain
RSSI Variation
147Sajid Hussain
RSSI Variation
Effect of Hill (small bump)
Same RSSI for 20 m and 25 m; however, LQI is different.
148Sajid Hussain
RSSI Variation
149Sajid Hussain
Effect of Discrete Power LevelsAt power level 3, the mote can easily communicate with motes within 5 meters as well as with motes within 10 meters. This behavior is also observed at other power levels such as power level 11. This is the lowest reliable power level for distances of 20 and 25 meters in our test environment. Since the power level setting is used to communicate to motes at both 20 and 25 meters away, the cost for transmitting over both these distances are also equivalent.While the costs for transmitting over 20 and 25 meters does not change, the quality of the signal is affected as one would expect.
150Sajid Hussain
Changing Packet Size
151Sajid Hussain
Effect of Packet SizeFrom our experiment we found that when transmitting between two nodes on the edge of the connected region, increasing the packet size to greater than 28 bytes decreases the PRR greatly. Further, even when in the connected region larger packet sizes have a negative effect on the link quality for those transmissions. From [1], as the packet size increases, the energy per bit of data decreases because the radio startup and overhead costs are shared amongst more data bits. With the results from our experiment we see that as the packet size increases the quality of the link decreases, resulting in a tradeoff between network link quality and transmission energy costs. One must choose a packet size that best matches the requirements of the network. For networks using links in the transitional region, we suggest that short packet sizes (28 bytes) would be suitable in order to avoid the large drop off in the packet reception rate.
[1] Q. Cao, T. He, L. Fang, T. Abdelzaher, J. Stankovic, and S. Son. Efficiency centric communication model for wireless sensor networks. In INFOCOM 2006: 25th IEEE International Conference on Computer Communications, pages 1–12, 2006.
152Sajid Hussain
Using Discrete Model
[EESR] S. Hussain and O. Islam, “An energy efficient spanning tree based multi-hop routing in wireless sensor networks,” in IEEE Wireless Communications and Networking Conference (WCNC), 2007.
[PEDAPPA] H. O. Tan and I. Korpeoglu, “Power efficient data gathering and aggregation in wireless sensor networks,” in SIGMOD Rec., 2003, pp. 66–71.
Latest Interests
154Sajid Hussain
Latest InterestsBeyond Layers
Cross Layer OptimizationCognitive NetworksCooperative Networks
SecurityConfidentiality, Privacy, Trust
Fault Tolerance and RobustnessScalability
155Sajid Hussain
New Interfaces
156Sajid Hussain
Merging of adjacent layersDesign two or more adjacent layers together superlayer is the union of the services provided by the constituent layersThis does not require any new interfaces
superlayer can be interfaced with the rest of the stack using the interfaces that already exist in the original architecture
157Sajid Hussain
Vertical calibration across layersRefers to adjusting parameters that span across layerBasically, the performance seen at the level of the application is a function of the parameters at all the layers below it
158Sajid Hussain
Implementing cross-layer interactionsDirect communication between layersA shared database across the layersCompletely new abstractions
159Sajid Hussain
Direct communication between layersA simple way to allow runtime information sharing between layers.The variables at one layer are visible at the other layers at runtime.
Direct Communication Between Layers
160Sajid Hussain
Shared database across the Layers
A common database that can be accessed by all layers
In one sense, the common database is like a new layer, providing the service of storage (retrieval) of information to all the layers.
Shared Database
161Sajid Hussain
Completely new abstractions
Allow rich interactions between the building blocks of the protocolsOffer great flexibility
both during design as well as at runtime
Require completely new system-level implementations
162Sajid Hussain
Cognitive and Cooperative Networks
163Sajid Hussain
Cognitive Network
164Sajid Hussain
Cognitive networkCognitive wireless access networks are those that can alter their operational parameters to respond to the needs of particular user/environment.
Observe
Analyze
Decide
Reconfigure
Measurement System
Handoff Adapters
Wireless transport /mobility management
Mac configuration
165Sajid Hussain
Virtual antenna arrayCo-operative diversityEnergy saving for larger distances
Multiple copies to the single/multiple receivers
166Sajid Hussain
Cui and Goldsmith [JSAC] ( Co-operative MIMO/ Virtual MIMO)
“Energy-efficiency of MIMO and Co-operative MIMO techniques in Sensor Networks” Shuguang Cui, Andrea J. Goldsmith, and Ahmad Bahai, IEEE JOURNAL IN SELECTED AREAS IN COMMUNICATIONS, Volume:22, Issue 6, August 2004
167Sajid Hussain
Virtual cross-layer-MIMO
1. Extended LEACH protocol.
2. Single receiver.3. Selection of appropriate
set of transmitter nodes.
4. Energy aware routing selection protocol
-Packet throughput-QoS
Virtual MIMO-based Cross Layer Design for Wireless Sensor Netoworks” Yong Yuan, Zhihai He, and Min Chen, Vehicular Technology, IEEE Transaction, vol.55, no. 3,
May2006
168Sajid Hussain
Cross-layer Co-operative MIMO(Cui and Goldsmith)
1. 2X2 co-operative MIMO2. Data link layer adjusts
with lower layers.3. Local information
exchange within cluster is not necessary.
4. Routing optimized considering each cluster as supernode.
“Cross-layer design of energy-constrained networks using co-operative MIMO techniques”, inivited for Publication at EURASIP’S Signal Processing Journal, August 2006
169Sajid Hussain
Virtual MIMO protocol architecture
1. CHs are selected randomly according to LEACH.
2. CH will select the number of transmitter and receivers.
3. Each cluster will have a energy consumption table for MIMO communication with nearby clusters.
4. Routing tree will be established by Prim’s Algorithm on the cost of energy consumption.
5. After the routing path is established, CH will announce the transmitter nodes, routing path, receiver nodes and optimal constellation size.
170Sajid Hussain
Different receiver sets
0.5J
0.4J
0.3J
BS
C1
C2
C3
Receiver node for C2
Cluster structure of C3
Receiver node for C1
Cluster Head
Transmitter node
171Sajid Hussain
WSN Test-bed (CBU)Radio Harsh EnvironmentsRemote AccessibilityNetwork ManagementExperimental ControlExtensible ScalabilityRepeatability InstrumentationData LoggingSignal AttenuationElectromagnetic Interference (EMI)
172Sajid Hussain
Hardware Architecture
S. Hussain, C. Ma, M. Murillo, J. Nicholson, N. Polu, and J. Slipp, “Winter: Architecture and applications of a wireless industrial sensor network testbed for radio-harsh environments,” in Proceedings of the Sixth IEEE/ACM International Conference on Communication Networks and Services Research (CNSR). IEEE Communication Society, May 2008.
173Sajid Hussain
Mote Platform
174Sajid Hussain
Software Architecture
175Sajid Hussain
Key Management in WSNsKey management is the process by which cryptographic keys are generated, stored, protected, transferred, loaded, used, and destroyed.The main concerns are as follows:
Key deployment /Pre-distributionKey establishment Node addition Node eviction
Sajid Hussain, Firdous Kausar, and Ashraf Masood, An Efficient Key Distribution Scheme for Heterogeneous Sensor Networks, in the Proceedings of the IEEE/ACM International Wireless Communications and Mobile Computing Conference 2007 (IWCMC 2007), Turtle Bay Resort Hotel, Honolulu, Hawaii, USA.
176Sajid Hussain
Basic Random Key DistributionKey Pre-distribution
Generate a large key pool S (e.g.,217-220 keys) and corresponding key identifiersAssigned a ring of m keys , drawn from the pool at random, without replacement.
Shared-key discoveryEach sensor node broadcasts a key identifier list and compares the list of identities received to the keys in their key ring.
Path Key EstablishmentCommunicate through intermediate nodes The intermediary node generates the path key and forwards it to the disconnected nodes through the shared keys generated in the previous phase
178Sajid Hussain
ReferencesBhaskar Krishnamachari, Networking Wireless Sensors, Cambridge Press, 2007.Chi-Yuan Chang, Chi-Yuan Chen, Chien-Hung Chen, Kai-Ti Chang, Diego Chung and Han-Chieh Chao, “Cross Layer issues for Heterogeneous networking in pervasive communications” Chapter IV of the Handbook on Mobile Ad Hoc and Pervasive Communications, to be published by the American Scientific Publishers.V. Srivastava and M. Motani, “Cross-Layer Design: A Survey and the Road Ahead,” IEEE Communications Magazine, vol. 43, issue 12, pp.112-119, Dec. 2005.V. Kawadia, P. R. Kumar, “A Cautionary Perspective on Cross-Layer Design,” IEEE Wireless Communications, vol. 12, issue 1, pp.3-11, Feb. 2005.