A Network of Wireless Sensor

Nodes
Sensing (Data Acquisition)

 Sensors capture phenomena in the physical world (process, system, plant)

 Signal conditioning prepare captured signals for further use (amplification, attenuation, filtering of unwanted
frequencies, etc.)

 Analog-to-digital conversion (ADC) translates analog signal into digital signal

 Digital signal is processed and output is often given (via digital-analog converter and signal conditioner) to an
actuator (device able to control the physical world)
Sensor Classification
 Power supply:

active sensors require external power, i.e., they emit energy (microwaves, light, sound) to trigger response or
detect change in energy of transmitted signal (e.g., electromagnetic proximity sensor)

passive sensors detect energy in the environment and derive their power from this energy input (e.g., passive
infrared sensor)

 Electrical phenomenon:
 resistive sensors use changes in electrical resistivity based on physical properties such as temperature
 capacitive sensors use changes in capacitor dimensions or permittivity based on physical properties
 inductive sensors rely on the principle of inductance (electromagnetic force is induced by fluctuating
current)
 piezoelectric sensors rely on materials (crystals, ceramics) that generate a displacement of charges in
response to mechanical deformation
Example: Wheatstone Bridge Circuit

 R1, R2, and R3 known (R2 adjustable)

 Rx is unknown

 Rx R2 
Vout  VCC    
R3  Rx R1  R2 
capacitive sensors
inductive sensors
WSN Communication

 Characteristics of typical WSN:

 low data rates
 energy-constrained sensors

 IEEE 802.11 family of standards

 most widely used WLAN protocols for wireless communications in general
 can be found in early sensor networks or sensors networks without stringent energy constraints

 IEEE 802.15.4 is an example for a protocol that has been designed specifically for short-range
communications in WSNs
 low data rates
 low power consumption
 widely used in academic and commercial WSN solutions
Single-Hop versus Multi-Hop

 Star topology:
 every sensor communicates directly (single-hop) with the base station
 may require large transmit powers and may be infeasible in large geographic areas
 Mesh topology
 sensors serve as relays (forwarders) for other sensor nodes (multi-hop)
 may reduce power consumption and allows for larger coverage
 introduces the problem of routing
Comparison
Traditional Networks Wireless Sensor Networks

General-purpose design; serving many applications Single-purpose design; serving one specific application

Typical primary design concerns are network performance and latencies; Energy is the main constraint in the design of all node and network
energy is not a primary concern components

Networks are designed and engineered according to plans Deployment, network structure, and resource use are often ad-hoc
(without planning)

Devices and networks operate in controlled and mild environments Sensor networks often operate in environments with harsh conditions

Maintenance and repair are common and networks are typically easy to Physical access to sensor nodes is often difficult or even impossible
access

Component failure is addressed through maintenance and repair Component failure is expected and addressed in the design of the
network

Obtaining global network knowledge is typically feasible and centralized Most decisions are made localized without the support of a central
management is possible manager
Node Architecture
 Analog-to-Digital Converter

 ADC converts the output of a sensor - which is a continuous, analog signal into a
digital signal. It requires two steps:
1. the analog signal has to be quantized
 allowable discrete values is influenced :
(a) by the frequency and magnitude of the signal
(b) by the available processing and storage resources
2. the sampling frequency
 Nyquist rate does not suffice because of noise and transmission error
 resolution of ADC - an expression of the number of bits that can be used to encode the digital output

 where Q is the resolution in volts per step (volts per output code); Epp is the peak-to-peak analog voltage; M is the ADC’s
resolution in bits
 The Processor Subsystem
 The processor subsystem
 its main purpose is to execute instructions related to sensing, communication, and self-organization

 It consists of
 processor chip
 nonvolatile memory - stores program instructions
 active memory - temporarily stores the sensed data
 internal clock

 The processor subsystem can be designed by employing one of the three basic
computer architectures
 Von Neumann architecture
 Harvard architecture
 Super-Harvard (SHARC) architecture
Von Neumann Architecture

 Von Neumann architecture

 provides a single memory space - storing program instructions and data

 provides a single bus - to transfer data between the processor and the memory
 Slow processing speed - each data transfer requires a separate clock
 Harvard Architecture
 Harvard architecture
 provides separate memory spaces - storing program instructions and data
 each memory space is interfaced with the processor with a separate data bus
 program instructions and data can be accessed at the same time
 a special single instruction, multiple data (SIMD) operation, a special arithmetic operation and a bit
reverse
 supports multi-tasking operating systems; but does not provide virtual memory protection
Super-Harvard Architecture
 an extension of the Harvard architecture
 adds two components to the Harvard architecture:
 internal instruction cache - temporarily store frequently used instructions - enhances performance
 an underutilized program memory can be used as a temporary relocation place for data
 Direct Memory Access (DMA)
Communication Interfaces

 Fast and energy efficient data transfer between the subsystems of a wireless
sensor node is vital
 however, the practical size of the node puts restriction on system buses
 communication via a parallel bus is faster than a serial transmission
 a parallel bus needs more space

 Therefore, considering the size of the node, parallel buses are never supported
 Serial Peripheral Interface
 SPI (Motorola, in the mid-80s)
 high-speed, full-duplex synchronous serial bus
 does not have an official standard, but use of the SPI interface should conform to the
implementation specification of others - correct communication
 The SPI bus defines four pins:
 MOSI (MasterOut/SlaveIn): used to transmit data from the master to the slave when a device is configured
as a master
 MISO (MasterIn/SlaveOut)
 SCLK (Serial Clock)
 used by the master to send the clock signal that is needed to synchronize transmission
 used by the slave to read this signal synchronize transmission
 CS (Chip Select) - communicate via the CS port
 Inter-Integrated Circuit
 Every device type that uses I2C must have a unique address that will be used to
communicate with a device
 In earlier versions, a 7 bit address was used, allowing 112 devices to be uniquely
addressed

 Currently I2C uses 10 bit addressing

 I2C is a multi-master half-duplex synchronous serial bus
 only two bidirectional lines: (unlike SPI, which uses four)
 Serial Clock (SCL)
 Serial Data (SDA)
 MAC Protocols in WSNs
 Most MAC protocols are built for fairness: everybody should get an equal amount of resources, no one should
receive special treatment
 In a WSN, all nodes cooperate to achieve a common purpose, therefore fairness is less of a concern
 Instead, wireless nodes are mostly concerned with energy consumption

 Reasons for energy inefficiency

 idle listening
 inefficient protocol designs (e.g., large packet headers)
 reliability features (collisions requiring retransmissions or other error control mechanisms)
 control messages to address the hidden-terminal problem
 overemitting
 Energy Efficiency
 Periodic traffic models are very common in WSNs
 significant energy savings can be obtained by putting a device into a low-power sleep mode
 fraction of time a sensor nodes spends in active mode is called the duty cycle
often very small due to the infrequent and brief data transmissions occurring in most sensor networks
 Contention-Free MAC Protocols
 Concept:
 allow only one sensor node to access the channel at any given time
 thereby avoiding collisions and message retransmissions

 Contention-free protocols allocate resources to individual nodes to ensure exclusive resource access by only one
node at any given time

 Exposes a number of desirable characteristics

 node knows exactly when it has to turn on its radio
 during all other times, radio can be turned off to preserve energy
 fixed slot allocations impose upper bounds on delay
 difficult to design schedules for large networks
 difficult to handle changes in topology, density, traffic load
 Traffic-Adaptive Medium Access
 TRAMA assumes a time-slotted channel, where time is dived into:
 periodic random-access intervals (signaling slots)
 scheduled-access intervals (transmission slots)
 Random-access intervals
 Neighbor Protocol (NP) is used to propagate one-hop neighbor information among neighboring nodes
nodes join a network by transmitting during a randomly selected slot
packets transmitted during these slots are used to gather neighborhood information by carrying a set of
added and deleted neighbors
in case no changes have occurred, these packets serve as “keep-alive” beacons
 NP allows nodes to obtain consistent two-hop topology information
 Traffic-Adaptive Medium Access
 Random-access intervals
 Schedule Exchange Protocol (SEP) establishes and broadcasts actual schedules (i.e., allocations of slots
to a node)
each node computes a duration SCHEDULE_INTERVAL
– represents the number of slots for which the node can announce its schedule to its neighbors
– this duration depends on the rate at which the node’s applications can produce packets
at time t, the node then computes the number of slots within [t, t+ SCHEDULE_INTERVAL] for which it
has the highest priority among its two-hop neighbors
the node announces the selected slots and the intended receivers using a schedule packet
the last slot in this schedule is used to announce the next schedule for the next interval
 Traffic-Adaptive Medium Access
 Random-access intervals
 Schedule Exchange Protocol (SEP)
example:
– a node’s SCHEDULE_INTERVAL is 100 slots
– the current time (slot number) is 1000
– a possible slot selection for interval [1000, 1100] for this node could be 1011, 1021, 1049, 1050, and
1093
– during slot 1093, the node broadcasts its new schedule for interval [1093, 1193]
 Traffic-Adaptive Medium Access
 Random-access intervals
 Schedule Exchange Protocol (SEP)
slot selection is based on the node’s priority at time t
– uses a pseudo-random hash of the concatenation of the node’s identity i and t:
prioi, t   hashi  t 
– node can indicate which slots it gives up, allowing other nodes to claim these unused slots
a node can determine its state for any given time slot t based on its two-hop neighborhood information and
the announced schedules
– node i is in the transmit (TX) state if it has the highest priority and if it has data to send
– node i is in the receive (RX) state if it is the intended receiver of the transmitter during slot t
– otherwise, the node can be switched into the sleep (SL) state
 Traffic-Adaptive Medium Access
 Summary
 compared to CSMA-based protocols
reduces the probability of collisions
increases the sleep time and energy savings
 unlike standard TDMA approaches
TRAMA divides time into random-access and scheduled-access intervals
during the random-access intervals
– nodes are awake to either transmit or receive topology information
– the length of the random-access interval affects the overall duty cycle and achievable energy savings of
a node
 Low-Energy Adaptive Clustering Hierarchy
 The LEACH protocol combines TDMA-style contention-free communication with a clustering algorithm for wireless
sensor networks
 a cluster consists of a single cluster head and any number of cluster members, which only communicate with
their cluster head
 clustering is a popular approach for sensor networks
facilitates data aggregation and in-network processing at the cluster head
reduces the amount of data that needs to be transmitted to the base station
 LEACH operates in rounds consisting of two phases:
 A setup phase phase
 A steady-state phase
 Low-Energy Adaptive Clustering Hierarchy
 Setup phase
 since the cluster head is responsible for coordinating cluster activity and forwarding data to the base station,
its energy requirements will be significantly larger compared to other sensor nodes
 therefore, LEACH rotates the cluster head responsibility among sensor nodes to evenly distribute the energy
load
every sensor i elects itself to be a cluster head with a certain probability Pi(t)

 There are various approaches to choose Pi (t)

 One of these approachs uses an indicator function Ci (t) to determine whether node i has been a cluster head in
the previous rounds
 Only nodes that have not been cluster heads recently are candidates for the cluster head role, in order to evenly
distribute the cluster head responsibility (and energy overhead) among all nodes
 Low-Energy Adaptive Clustering Hierarchy
 This approach does not consider the actual amount of energy available to each node
 An alternative approach to determining the probability of becoming a cluster head can be used:

 Ei t  
Pi t   min  k ,1
 Etotal t  
 Ei(t) is node i’s actual current energy and Etotal(t) is the sum of the energy levels of all nodes
 a disadvantage of this approach is that every node must know (or estimate) Etotal(t)
 Low-Energy Adaptive Clustering Hierarchy
 Once a sensor node has determined that it will serve as cluster head for the next round, it informs other sensor
nodes of its new role by broadcasting an advertisement message (ADV) using a non-persistent CSMA protocol

 Every sensor node joins a cluster by selecting the cluster head that can be reached with the smallest amount of
transmit energy, based on received signal strength of the ADV messages from the cluster heads

 Joining is achieved by transmitting a join-request (Join-REQ) message to the chosen cluster head using CSMA

 The cluster head establishes a transmission schedule for its cluster and transmits this schedule to each node in its
cluster
 Steady-state phase
 a sensor node communicates only with the cluster head, it is allowed to transmit data only during its allocated
slots (indicated by the schedule received from the cluster head)
 the responsibility of the cluster head is to forward sensor data originating at one of its sensor nodes to the base
station
 in order to preserve energy each cluster member turns off the wireless radio between its designated slots
 the cluster head remains awake at all times to receive sensor data from its cluster members and to
communicate with the base station
 Low-Energy Adaptive Clustering Hierarchy
 LEACH-C
 variation of LEACH protocol that relies on the base station to determine the cluster heads
during setup phase, each sensor node transmits its location and energy levels to the base station
based on this information, the base station determines the cluster heads and informs the cluster heads of
their new role
other sensor nodes can then join clusters using join messages as described in the original LEACH protocol
Classification of Routing Protocols
 Classification of Routing Protocols
 Network organization
 flat: all nodes are “equal”
 hierarchical: different “roles” for different nodes (e.g., cluster heads versus cluster members)
 location-based: nodes rely on location information
 Route discovery
 reactive (on-demand): find route only when needed
 proactive (table-driven): establish routes before they are needed
 hybrid: protocols with reactive and proactive characteristics
 Protocol operation
 negotiation-based: negotiate data transfers before they occur
 multi-path: use multiple routes simultaneously
 query-based: receiver-initiated
 QoS-based: satisfy certain QoS (Quality-of-Service) constraints
 coherent-based: perform only minimum amount of in-network processing
 Routing Metrics (Example)

 Numbers along links: cost for transmission over link

 Numbers in parentheses: remaining energy capacity
 SPIN Family of Protocols
 Sensor Protocols for Information via Negotiation (SPIN)
 Example of data-centric routing
 Uses negotiations to address all problems of flooding
 implosion: nodes negotiate before data transmission
 overlap: nodes negotiate before data transmission
 resource blindness: resource manager keeps track of actual resource consumption and adapts routing and
communication behavior

 SPIN uses meta-data to succinctly and completely describe sensor data

 Requirements:
 if x describes the meta-data for some sensor data X, the size of x (in bytes) must be less than the size of X
 if two pieces of sensor data differ, their meta-data representations should differ too
SPIN-PP
 Only nodes needed the advertised data respond
 Nodes can aggregate received data with their own and advertise aggregate
 Nodes need to know their single-hop neighbors only
 Lost ADV messages: periodically re-advertise
 Lost REQ/DATA messages: re-request data of interest if data does not arrive within certain timeout interval

 SPIN-EC:
 adds energy conservation as long as energy sufficient, node participates in 3-way handshake
 nodes does not participate if it believes that this will reduce its energy below a certain low-energy threshold
 node replies to ADV only if sufficient energy for transmitting REQ and receiving DATA
 node initiates handshake only if it has sufficient energy to send DATA to all neighbors
 SPIN-BC:
 uses one-to-many communications (broadcast)
 receiver node waits for a random time interval before issuing REQ; if other node’s REQ overheard, the receiver
node cancels timer and does not send its own REQ
 advertiser broadcasts DATA only once (ignore duplicate REQs)
 Location-Based Routing
 One single destination
 Each forwarding node makes localized decision based on the location of the destination and the node’s neighbors
(greedy forwarding)
 Challenge: packet may arrive at a node without neighbors that could bring packet closer to the destination (voids
or holes)
 Forwarding Strategies
 Greedy: minimize distance to destination in each hop (node E)
 Nearest with Forwarding Progress (NFP): nearest of all neighbors that make positive progress (in terms of
geographic distance) toward destination (A)
 Most Forwarding Progress within Radius (MFR): neighbor that makes greatest positive progress (progress is
distance between source and its neighbor node projected onto a line drawn from source to destination) (B)
 Compass Routing: neighbor with smallest angle between a line drawn from source to the neighbor and the line
connecting source and destination (C)