Transcript Template - Elsevier

HANDBOOK ON GREEN INFORMATION AND
COMMUNICATION SYSTEMS
Chapter 4:
IEEE 802.15.4 Based Wireless Sensor
Network Design for Smart Grid
Communications
Chun-Hao Lo and Nirwan Ansari
Advanced Networking Laboratory
Department of Electrical & Computer Engineering
New Jersey Institute of Technology
Newark, New Jersey, USA
Agenda
 Wireless sensor networks (WSN) and associated
applications supported in Smart Grid Communications
 A comparison of IEEE 802.15.4 and Power Line
Communications technologies
 An introduction of IEEE 802.15 Task Groups,
particularly the IEEE 802.15.4g Task Group (TG4g) in
developing the Smart Utility Network/Neighborhood
(SUN) design
 Discussion of studies and challenges in IEEE 802.15.4
LR-WPAN with respect to network design in PHY/MAC
layers, fairness, routing, and security/privacy issues
 Conclusions
2
WSNs in Smart Grid Communications (1/4)
Wireless Sensor Networks (WSNs) are deployed
throughout the electric power system from
generation, transmission, distribution, to end-use
sectors
Applications: equipment sensing and monitoring,
fault diagnosis, meter reading, etc.
Components: Supervisory Control and Data
Acquisition and Energy Management Systems
(SCADA/EMS), Phasor Management Units and
Phasor Data Concentrators (PMU/PDC), Advanced
Metering Infrastructure (AMI), and a wide range
of Remote Terminal Units (RTUs), etc.
3
WSNs in Smart Grid Communications (2/4)
4
WSNs in Smart Grid Communications (3/4)
Five major domains
 Traditional power plants, transformers and substations
control, Distributed Energy Resources (DERs), power
lines monitoring, and demand-side customers
5
WSNs in Smart Grid Communications (4/4)
Types of sensors
 chemical, electrical, environmental, pressure, smart
appliances sensors, smart meters, etc.
Different classes of sensor data to meet different
latency requirements, e.g., voltage and frequency
control (< 100ms), smart metering (> 1s)
Collected data may be shared and reused for
multiple applications (Ref. [9])
 Challenges: modification to data packet headers may be
required; data may not carry sufficient information for
some specific applications
 Developments: Advanced sensors and associated sensor
data management
6
IEEE 802.15.4 vs. PLC Technology
IEEE 802.15.4 (Ref. [17][18])
 Fast deployment, low implementation cost, low complexity, low energy
consumption
 Matured technology used in various applications and tailored by
popular working groups Alliances, e.g., ZigBee, WirelessHART,
ISA100
Power line Communications (PLC) (Ref. [2][12][58])
 Another viable approach that utilizes existing power line cables as
the communications medium for data transmission
 Shortcomings: 1) high bit error rates (due to noisy power line, e.g.,
motors, power supplies), 2) limited capacity (attributed to the
number of concurrent network users and applications concurrently
being used), 3) high signal attenuation (dependent of geographical
locations), 4) phase change between indoor and outdoor
environments, and 5) disconnected communications due to opened
circuits
7
IEEE 802.15 Task Groups
8
PHY specifications in IEEE 802.15.4a, b, c, and d
 The legacy IEEE 802.15.4 standards adopt BPSK, ASK, O-QPSK
modulations, support data rates 20, 40, 100, 250 kbps, and operate in
868/915 MHz and 2.4 GHz frequency bands
 The standards only specify PHY and MAC layers and leave the upper
layers to be designed by the application designers
9
IEEE 802.15.4g Task Group (TG4g)
 TG4g specifies Smart Utility Network/Neighborhood (SUN)
development by tackling a number of technical challenges in
communications systems for the utility operators, especially
the interference and coexistence issues
 TG4g amends the legacy IEEE 802.15.4 standard for SUN PHYs while IEEE
802.15.4e is tailored for SUN MAC (Ref. [59])
 Three major SUN PHYs are proposed: multi-rate/multi-regional frequency
shift keying (MR-FSK), multi-rate orthogonal frequency division multiplexing
(MR-OFDM), and multi-rate offset quadrature phase shift keying (MROQPSK) (Ref. [59])
 Bands allocated in domains/countries for SUN are 470–510 MHz (China),
863–870 MHz (Europe), 902–928 MHz (United States), 950–958 MHz
(Japan), and 2.4–2.4835 GHz (worldwide)
 Keys: utilization of sub-GHz frequency bands (i.e., licenseexempt bands below 1 GHz), and development of multi-PHY
management (MPM) (Ref. [60])
10
Other Approaches
Other techniques have been proposed to enhance
the network performance in smart grid
communications from the PHY perspective




Multi-channel access (Ref. [14])
WiFi features adoption (Ref. [15])
Cognitive radio (Ref. [16])
TV White Space (Ref. [59])
11
IEEE 802.15.4 Studies and Challenges (1/4)
LR-WPAN generally employs TDMA with CSMA-CA,
and adopts DSSS for various modulation schemes
Network variables and metrics in LR-WPAN design
are predominantly based on topology control and
traffic engineering




Network size
Node placement
Data packet size
Traffic loads
12
IEEE 802.15.4 Studies and Challenges (2/4)
The network performance of LR-WPAN is
determined by several key factors
 Frequency of wireless medium contention
 Successful data delivery ratio; collisions from hidden
node transmission, congestions from heavy traffic loads,
and packet losses and drops from wireless deterioration
and buffer overflow
 Latency; unnecessary delayed transmission from the
exposed node problem, a clumsy increase in MAC CSMA
backoff periods, and inflexible routing design
 Energy depletion rate; affected by the duty-cycle
arrangement as well as data aggregation and fusion
mechanisms.
13
IEEE 802.15.4 Studies and Challenges (3/4)
Wireless impairments such as background noise,
signal attenuation, path loss, multipath/fading, and
interference are also found in LR-WPAN
 Several measurements and parameters specified in
LR-WPAN PHY/MAC are principal attributes to the
network performance and design
 Receiver energy detection (ED) within the current channel
 Link quality indicator (LQI) for received packets and channel
frequency selection
 Clear channel assessment (CCA) for CSMA-CA
 NB: the number of times that CSMA-CA is required to backoff
 BE: a backoff exponent that is used to calculate the backoff period
 CW: the contention window length
14
IEEE 802.15.4 Studies and Challenges (4/4)
 The PHY payload in IEEE 802.15.4 is limited to 127 bytes;
the application payload (useful information) is reduced to 60
bytes ~ 80 bytes after an inclusion of control bits. Since
the ratio of overhead to data payload is considerably large,
one needs to determine
 How to use bandwidths in LR-WPAN efficiently?
 How to manage packet size with useful data to achieve low delay and
low packet-loss rate during transmission? (Ref. [38])
 Two types of data packet collision can also be found in LRWPAN (RTS/CTS is not supported in IEEE 802.15.4)
 Collision due to regular medium contention
 Collision due to hidden node problem (Ref. [34][35][37])
 Exposed node problem can occur in LR-WPAN as well (Ref.
[36])
15
IEEE 802.15.4 Superframe structure (1/2)
 Two operation modes
CAP – Contention access period
CFP – Contention free period
BSD – Base slot duration
SD – Superframe duration
BSFD – Base superframe duration
NSFS – Number of superframe slots
BI – Beacon interval
SO – Superframe order
BO – Beacon order
 beacon-enabled (B-E) with slotted CSMA-CA mode, and beaconless (BL, i.e.,
beacon-disabled) with unslotted CSMA-CA mode
 In the B-E mode, the superframe is bounded by two consecutive
beacons, and constructed by the active and inactive parts
 The active portion is divided into 16 equal time slots that comprises CAP
and CFP, which defines GTS
 Up to 7 GTSs can be allocated by a WPAN coordinator and each GTS
may occupy more than one slot period (i.e.,  1 BSD)
16
IEEE 802.15.4 Superframe structure (2/2)
CAP – Contention access period
CFP – Contention free period
BSD – Base slot duration
SD – Superframe duration
BSFD – Base superframe duration
NSFS – Number of superframe slots
BI – Beacon interval
SO – Superframe order
BO – Beacon order
 GTS allocation and management specify starting slot, length, direction
(i.e., transmit or receive), and associated node address. Each GTS is
allocated first come first serve and released when it is not required
 Slot boundary rule: a node begins to transmit on the next available slot
boundary when the channel is idle. Otherwise, it allocates the boundary
of the next backoff slot before it goes into the backoff stage. If the
time between the next available backoff slot and the end of the active
period is not long enough for a node to complete its transmission, it may
have to wait until the arrival of the next superframe
17
Network design for IEEE802.15.4-based WSN (1/9)
A number of principal research issues in IEEE
802.15.4 are categorized into four areas: PHY/MAC
layers, fairness, routing, and security
 Analysis in PHY/MAC
under different network
environments is grouped
into B-E and BL studies
 In B-E study, CAP/CFP and
BO/SO are examined
 In both studies, ED/LQI,
CCA, CC/HNC/ENP, and
NB/BE/CW are
investigated
18
Network design for IEEE802.15.4-based WSN (2/9)
CAP and CFP (with GTS) Management
 QoS consideration in data transmission specified in smart
grid applications, e.g., GTSs are allocated to nodes with
mission-critical data
 The positions of CAP and CFP are swapped (modification
to the standard is required) in order to grant the
retransmission attempt of GTS to proceed in CAP of the
same superframe upon a failed transmission in GTS (Ref.
[42])
 Analysis of GTS request drop due to possible collisions in
CAP when BO is considerably small (Ref. [43])
 Two-traffic class is proposed to allow nodes with higherpriority data to transmit by assigning CW=1 (Ref. [44])
19
Network design for IEEE802.15.4-based WSN (3/9)
SO and BO Measurement
 Consideration of the need for power saving on each node
at the cost of transmission latency, i.e., SO=BO (100%
duty cycle) if a node is not power-constrained
 Analysis of end-to-end delay and packet loss by studying
the packet inter-arrival time and the ratio of BO to SO
(Ref. [45])
 Tradeoff between latency and energy consumption under
the same duty cycle, which can be constructed by
different combination sets (Ref. [46]), e.g., both
BO=3/SO=2 and BO=11/SO=10 cases have 50% duty cycle
1
Duty cycle   
2
BO  SO
, 0   BO  SO   14
20
Network design for IEEE802.15.4-based WSN (4/9)
ED and LQI Assessment (Ref. [47])
 Determination of ED and LQI to identify the radio
condition
 While using LQI and RSSI (or ED) metrics for a number
of field tests in real-world power delivery and
distribution systems, several conclusions are made: 1) the
background noise (varied in temperature and time) is
higher for the indoor than the outdoor environment; 2)
channel 26 in IEEE 802.15.4 is not influenced by IEEE
802.11b interference; and 3) LQI is a good estimator
when the signal is found below and close to the sensitivity
threshold, i.e., -94 dBm; otherwise, RSSI (or ED) is
recommended
RSSI – Received signal strength indicator
21
Network design for IEEE802.15.4-based WSN (5/9)
CCA Analysis
 Determination of whether a specific radio channel is busy
or idle prior to the data transmission
 Collision may occur during the receive-to-transmit (Rxto-Tx and vice versa) turnaround time even if a channel
was initially detected as idle (Ref. [48])
 An adaptive MAC engine containing a collection of preset
optimal protocols for different network conditions is
proposed to avoid time spent on restarting the design
process each time (Ref. [49])
22
Network design for IEEE802.15.4-based WSN (6/9)
NB, BE, and CW Examination (Ref. [50])
 NB and BE parameters can be directly affected in
consequence of CCA, which is related to CW assignment
 Under light or medium traffic condition, increasing the
BE value seems to bring down the probability of packet
loss, however, at the cost of increased latency
 Under heavy traffic condition, adjusting BE becomes
insignificant to improve network performance
23
Network design for IEEE802.15.4-based WSN (7/9)
Fairness
 An adaptive GTS allocation scheme is proposed to
determine the success of GTS requests and the present
traffic-level state of a node (Ref. [51])
 A node generating heavy or more recent data traffic is
likely to have a higher probability of staying in a higher
priority state
 A node staying in a higher-level state with temporary
transmission interruption will slightly be demoted to a
lower state. On the other hand, a node in a lower-level
state can be promoted to a higher state if a consecutive
success of GTS requests is achieved
24
Network design for IEEE802.15.4-based WSN (8/9)
Routing Arrangement
 While the standard does not specify network/transport
layer, various routing protocols based on AODV have been
proposed (Ref. [52][53][54])
 A routing strategy based on OLSR that responds to the
requirements specified in power generation industry is
also proposed (Ref. [10])
 A hybrid routing scheme unifying flat and hierarchical
multi-hop algorithms with respect to power consumption
is also proposed (Ref. [33])
 New integrated routing techniques in supporting IPv6 via
6LoWPAN need to be developed (Ref. [55])
25
Network design for IEEE802.15.4-based WSN (9/9)
Security and Privacy
 Owing to the low computation capability and high
overhead constraints, limit of number of access control
list (ACL) entries and lack of group keying are identified
(Ref. [56])
 Security architecture for smart grid WSNs specifying
security standards and testing/evaluation for both
hardware and software need to be developed (Ref. [13])
 Privacy in smart grid communications is comparable to
patients' medical records in home and hospital
 Elliptic curve cryptography adopted in healthcare WSN
is proven to be lightweight computationally and uses
smaller key sizes for obtaining the same security level as
compared to RSA (Ref. [57])
26
A Summary of Network design and challenges in
IEEE802.15.4-based WSN
27
Conclusions (1/2)
 Smart grid applications with different bandwidth and latency
requirements can be provisioned in HR-WPAN (IEEE 802.15.3
based) and LR-WPAN (IEEE 802.15.4 based), which require
further investigations for smart grid communications
(improvement to legacy IEEE 802.15.4)
 Design of data prioritization related to specific applications and QoS
requirements
 Adequacy of control (i.e., overheads) and data packet size (including
commands)
 Schemes for multi-PHY management
 Assessment of communications link quality
 Innovation of MAC medium contention
 Flexibility of routing mechanisms
 Fairness issues upon adopted schemes
 Security/privacy models for protecting data and associated
transmission
28
Conclusions (2/2)
 Proposed techniques to alleviate interference and
coexistence problems by utilizing spectrums more
effectively and efficiently, e.g., operating frequency bands
below 1 GHz and developing multi-PHY management (specified
in IEEE 802.15.4g), and adopting TV White space, WiFi
features, multi-channel access, as well as cognitive radio in the
legacy IEEE 802.15.4 standard (spectrum use efficiency)
 Complementary strategy of combining IEEE 802.15.4 with
PLC technologies should be considered to provision sensor
applications in various smart grid domains (interoperability)
 Innovative mechanisms and models of integrating IP and
other technologies with WSNs need to be developed to
facilitate smart grid communications and management
(integration)
29
Thanks for your
attention!
30