Sie sind auf Seite 1von 6

DYMO as routing protocol for IEEE-802.15.

4 Enabled Wireless Sensor Networks


A. S. Raghuvanshi S. Tiwari
Indian Institute of Information Technology, IIITA Motilal Nehru National Institute of Technology, MNNIT
Deoghat, Jhalwa Teliar Ganj
Allahabad 211012 Allahabad 211004
India India
asr@iiita.ac.in stiwari@mnnit.ac.in

Abstract- Wireless Sensor Networks (WSNs) is an emerging Recently adopted IEEE 802.15.4 standards for Low-rate
field of engineering with various applications. They have Wireless Personal Area Network (LR-WPAN)[8], which
emerged as a new class of large scale networks of embedded has been designed to have properties of low cost, low data
systems with limited communication, computation and energy rate, short transmission range and most importantly low
resources. Recently adopted IEEE 802.15.4 standards for Low-
power consumption makes it suitable for WSNs. IEEE
rate Wireless Personal Area Network (LR-WPAN) are most
extensively used communication platform for WSNs. 802.15.4 standards focuses only on physical (PHY) and
Conventional routing in IEEE 802.15.4 networks is based on medium access control (MAC) layer upper layer of the
Ad hoc On-demand Distance Vector (AODV) which facilitates protocol are kept user specific. ZigBee Alliance [9] which
the building of mesh networks for data dissemination. The deals with marketing, interoperability, and commonly
Dynamic MANET On-demand (DYMO) routing protocol is a accepted specifications for WPAN devices manufactured
simple and fast routing protocol which is still in its internet by different manufacturers. ZigBee has devices a
draft and is in its 21st version. It is designed basically for multi- specification [10] for higher layer communication protocol
hop ad-hoc networks. In this paper we present a performance suite based on 802.15.4 PHY and MAC specifications.
evaluation of DYMO routing protocols for IEEE-802.15.4
Most of the commercially available WSN platforms such
enabled WSNs. The performance of routing protocols is carried
out base upon average energy consumption, average as MICA, MICAZ, Imote2, TeslosB, TI CC243 etc have
throughput, average end to end delay and average battery adopted IEEE 802.15.4 standard for their PHY and MAC
consumption and is compared with presently used AODV. specifications and ZeegBee for their upper layer
I INTRODUCTION communication protocols specifications. Routing of data in
ZigBee is based on the well-known ‘Ad-Hoc On-demand
Wireless Sensor Networks (WSN) can be defined as a self Distance Vector’ (AODV) protocol [11]. AODV is a
configured network formed using a large numbers of small reactive protocol that was basically designed for highly
embedded devices, each with sensing, computation and dynamic wireless mobile ad hoc networks. AODV is
communication capabilities. Each device is called sensor capable of both unicast and multicast routing in Ad hoc
node that are distributed over a geographical area with networks. It is an on demand algorithm, meaning that it
either predefined locations or randomly deployed. builds routes between nodes only as desired by source
Individual sensor nodes posses limited processing, nodes. Dynamic On-demand MANET routing protocol
communicating and energy recourses. These sensor nodes (DYMO)[12] is the current engineering focus for reactive
can sense, measure, and gather information from the routing in the MANET. DYMO, operates very similarly to
vicinity, based the local pre defined decision process, can AODV, but requires only the most basic route discovery
transmit the sensed information (data) to the user or and maintenance procedures only. In this study we present
information sink [1]-[3]. Recent studies [4], [5] shows that a performance analysis of DYMO as compare to AODV
the employment of WSNs for industrial applications is for IEEE 802.14.5 enabled WSNs. Section II presents an
expected to increase at an exponential pace in coming overview of adopted IEEE 802.15.4 standard for their PHY
years with their intrusion in the fields of logistics, and MAC. Section III presents a brief description of
automation and control. Sensor nodes work on batteries AODV and DYMO routing. Section IV presents a brief
and most of times it may be impossible to replace or description of chosen metric for result comparison by
recharge the batteries once network is deployed. In some simulation, section V presents results of our study. Section
cases it may be possible to scavenge energy from external VI concludes the study.
environment to recharge the primary energy source [6]. In
any case energy in WSN is a constrained resource and IEEE 802.15.4 STANDARD OVERVIEW
must be utilized judiciously to improve the lifetime of The IEEE 802.15.4 standard [8] which defines the physical
Sensor network .Energy consumption for communication (PHY) and medium access layers (MAC) for low cost, low
and computational processing is well studied and research rate personal area networks (LR-WPAN). The advantages
shows that communicating 1 kB of data over 100m of an LR-WPAN are ease of installation, reliable data
distance consumes roughly the same amount of energy as transfer, short range of operation, extremely low cost,
computing 3 million instructions [7]. Communication of energy efficient duty cycling for long battery life, while
sensed data from the sensor nodes happens to be main maintaining a simple and flexible protocol stack for upper
cause of energy depletion for sensor nodes. In order to layers. The energy efficient duty cycling makes it most
extend the active life of WSN we need to process data promising technology for WSNs.
locally and send the processed information to sink node. IEEE 802.15.4: The physical layer (PHY)
One of possible solution for optimal use of available The physical layer supports three frequency bands: a 2450
energy is by using energy efficient routing protocols for MHz band (with 16 channels), a 915 MHz band (with 10
data dissemination. channels) and a 868 MHz band (1 channel), all using the

978-1-4244-9730-0/10/$26.00 ©2010 IEEE


Direct Sequence Spread Spectrum (DSSS) access mode. (GTS) assigned and administered by the PAN coordinator.
The 2450 MHz band employs Offset Quadrature Phase When an end-device needs to send data to a coordinator
Shift Keying (O-QPSK) for modulation while the 868/915 (non GTS) it must wait for the beacon to synchronize and
MHz bands rely on Binary Phase Shift Keying (BPSK). then contend for channel access. On the other hand,
Table 1 summarizes the main features of the three bands. communication from a coordinator to an end-device is
Besides activation and deactivation of the radio indirect. The coordinator stores the message and
transceiver, the physical layer supports functionalities for announces pending delivery in the beacon. End-devices
channel selection, link quality estimation (LQI) for usually sleep most of the time and wake up periodically to
received packets, energy detection (ED) measurement for see if they have to receive some messages from the
current channel and clear channel assessment (CCA) for coordinator by waiting for the beacon. When they notice
CSMA-CA Channel frequency selection for data that a message is available, they request it explicitly during
transmission and reception. the CAP. When a coordinator wishes to talk to another
Table 1 Physical Layer Specification for IEEE 802.15.4 coordinator it must synchronize with beacon similar to an
Properties 2450 MHz 915 MHz 868 MHz end-device. The other option for PAN communication is to
Gross Data Rate 250kbps 40kbps 20kbps do without a Super-frame. The PAN coordinator never
Number of channels 16 10 1
Modulation O-QPSK BPSK BPSK
sends beacons and communication happens on the basis of
Pseudo noise chip 32 15 15 un-slotted CSMA-CA. But in this case the coordinator has
sequence to be always in active mode, ready to receive data from
Bits per symbol 4 1 1 end-devices. Data transfer in the opposite direction is
Symbol period 16µs 24 µs 49 µs
>15ms >15ms >15ms
based on polling where; the end device periodically wakes
Latency
Range 10-20m 10-20m 10-20m up and polls the coordinator for pending messages. The
coordinator sends either the available messages or the
IEEE 802.15.4: The MAC layer signal that none is available. In this case coordinator nodes
are active all the time hence communication between
The MAC layer of IEEE 802.15.4 standard defines two coordinators poses no problems. In addition to data
types of nodes, Reduced Function Devices (RFDs) and transfer, the MAC layer offers channel scan and
Full Function Devices (FFDs). FFDs are equipped with a association/disassociation functionalities.
full set of MAC layer functionalities, which enables them
to act both as a network coordinator or a network end-
device. When acting as a network coordinator, FFDs send
beacons that provide synchronization, communication and
network join services to other devices. RFDs can only act
as end-devices and are equipped with sensors for detection
or measurements. They may only interact with a single
FFD to route the sensed information to the sink using
multi-hop communication. Depending upon the
application, IEEE 802.15.4 may operate either of the two
main types of network topologies, namely, the star Figure No.1 Super frame structure for Beacon enabled MAC IEEE
topology and the peer-to-peer topology. In the star 802.15.4
topology, a master–slave network model is adopted where
an FFD takes up the role of PAN coordinator the other IEEE 802.15.4 LOW-POWER OPERATION
nodes can be either RFDs or FFDs that will only
communicate to the PAN coordinator. In the peer-to-peer IEEE 802.15.4 MAC layer provides a duty-cycle operation
topology, a FFD can talk to other FFDs within its radio by setting two system parameters, macBeaconOrder(BO)
range and can relay messages to other FFDs outside of its and macSuperframeOrder(SO), such that low power
radio coverage through an intermediate FFD, forming a consumption is achieved for the deployed networks [8]. In
multi-hop network. One of the PAN coordinator (normally addition to determining the duty cycle, the BO and SO
first one) is selected as network coordinator in order to parameters collaboratively provide various low-power
administer network operation. The PAN coordinator may attributes under a single duty cycle for diverse
operate its PAN with a Super-frame or without it environments [13]. WSN are energy constrained system
depending upon the battery conservation need. In the first hence low power operation of IEEE 802.15.4, needs
case it starts the Super-frame with a beacon serving for special mention. According to the IEEE 802.15.4 standard
synchronization and to describe the Super-frame structure the super-frame active period is called super-frame
for sending control information to the PAN. As shown in duration (SD) and Beacon Interval (BI) for given BO and
Figure No. 1, the Super-frame is divided into an active SO are defined as follows
period called Super-frame Duration (SD) and an inactive 2 (1)
period, where the PAN coordinator may go to sleep for 2 (2)
conserving battery. The active portion is further divided Where
into fixed size slots containing a Contention Access Period
(CAP), and a Contention Free Period (CFP). During CAP,
nodes compete for channel access using a slotted CSMA-
CA protocol, and during CFP nodes transmit without
contending for the channel, in Guaranteed Time Slots
operation.
0.960 16 AODV
With Ad-Hoc On-demand Distance Vector’ (AODV) protocol
0 (3) [15] is a reactive protocol that was basically designed for
For beacon enable operation, 0 14 and 15 , highly dynamic wireless mobile ad hoc networks. It is
for non beacon enable operation. adopted as standard routing protocol for ZigBee
Low power operation for network is achieved by achieving specification [10] To find a route to the destination, the
low duty-cycle. The duty cycle in IEEE 802.15.4 is defined source node that does not have destination route,
as the proportion of time during which a system is active broadcast Route request packets (RREQs)to neighboring
(SD) with respect to beacon interval (BI) and can be nodes, which then forwards the request to their neighbors
calculated as. and so on. In order to control network-wide broadcasts of
1 RREQ packets, the source node uses an expanding ring
100% 100% search technique where source node starts searching the
2
destination using some initial time to live (TTL) value. If
no reply is received within the discovery period, TTL
value incremented by a pre defined increment value. This
process continues until a predefined threshold is reached.
When an intermediate node forwards the RREQ, it records
the address of the neighbor from which first packet of the
broadcast is received, thereby establishing a reverse path.
When the RREQ reaches a node that is either the
destination node or an intermediate node with a fresh
enough route to the destination, replies by uni-casting the
route reply (RREP) towards the source node. As the RREP
is routed back along the reverse path, intermediate nodes
along this path set up forward path entries to the
Figure No. 2 Super-frame structure for different (BO, SO) settings for the destination in its route table and when the RREP reaches
duty cycle of 50%
the source node, a route from source to the destination
establish. A route established between source and
When the duty cycle ≤ 100%, devices can turn off their
destination pair is maintained as long as needed by the
transceivers during inactive periods to reduce power source. If the source node moves during an active session,
consumption. A large duty cycle results in higher power route discovery is reinitiated to find a new route to
consumption as devices remain in active state for longer destination. However, if the destination or some
periods. Additionally, a specific duty cycle can be intermediate node moves, the node upstream of the break
achieved by setting several different combinations of (BO, remove the routing entry and send route error (RERR)
SO) pairs. For example, the (BO, SO) pairs, (5, 3), (6, 4), message to the affected active upstream neighbors. These
and (7, 5) have the same duty cycle 1/4 (i.e., 25%). nodes in turn propagate the RERR to their precursor nodes,
However, different pairs under the same duty cycle provide and so on until the source node is reached. The affected
various low power attributes, and have differing impacts source node may then choose to either stop sending data or
on throughput and energy efficiency. Figure No.2 presents reinitiate route discovery for that destination by sending
two examples of a super-frame structure for different (BO, out a new RREQ message. AODV is designed for highly
SO) settings for the duty cycle of 50%. It can be seen if a dynamic mobile networks but the un-predictable topology
large BO is used, inactive periods are extended and change in WSN due to node failure makes them virtual
transmission latency will increase. Also, with a long mobile networks. Hence a reactive routing protocol is a
inactive period, more buffered packets will contend for good choice especially for event driven or periodic data
channel simultaneously at the start of the next active driven WSN applications.
period. This will result in throughput degradation due to DYMO
increased packet drop rate. Conversely, if BO is small, the The Dynamic MANET On-demand (DYMO) [12] routing
length of the super-frame (BI) decreases, but the beacon protocol is a simple and fast routing protocol for multi-hop
transmission frequency will increase. As number of networks. It discovers uni-cast routes among DYMO
beacons increase, available bandwidth for data delivery routers within the network in an on-demand fashion,
will decrease. Furthermore, overall power consumption offering improved convergence in dynamic topologies.
increases due to increase in beacon transmissions. When a source needs to send a data packet, it sends an
However, a small BO benefits the throughput since the RREQ to discover a route to the destination same as
number of buffered packet will be less due to a shortened AODV. After issuing an RREQ, the origin DYMO router
inactive period. If the duty cycle is extremely small, the waits for a route to be discovered. If a route is not obtained
throughput could be reduced significantly because of within RREQ waiting period, it may again try to discover a
insufficient bandwidth. Therefore, a trade-off exists for route by issuing another RREQ. To reduce congestion in a
duty-cycle setting between system throughput and device network by repeated attempts for route discovery it utilizes
energy consumption. For WSN energy consumption is one exponential backoff mechanism. Data packets awaiting to
of the most important parameter, hence proper selection of be routed are buffered in a predefined sized buffer that
BO and SO is very important. [14] Presents a discards older packets first.. Buffering of data packets can
comprehensive analysis of IEEE 802.15.4 duty cycle have either positive or negative effects on the system
performance. Therefore buffer size and settings should be Average energy consumption
administratively configured or intelligently controlled. If a
route discovery has been attempted for a predefined It indicates the average amount of energy spent in each
maximum attempts and a route to the target node is not node for each individual task. An increased value of
discovered, all data packets intended for that node are average energy consumption indicates more power
dropped from the buffer and a Destination Unreachable consumption by each node. This metric is computed with
ICMP message is delivered to the source. When a data the help of the following equation
packet is to be forwarded and cannot be delivered due to ∑ (TE + REi + IEi )
N
i
missing route to the Destination an RERR is issued. An Avr Energy Consumption = 1

RERR notifies DYMO routers that a link break has N


occurred and certain routes are no longer available. If the where for ith node
route with the broken link has not been used recently, the TEi = Energy Consumed for transmission of packet for node i
RERR should not be generated. DYMO is applicable to
REi = Energy Consumed for reception of packet for node i
memory constrained devices, since little routing state is
maintained in each DYMO router. Only routing IEi = Energy Consumed in idle ( active with no data) for node i
information related to active sources and destinations is N = Number of Nodes
maintained. WSN being a resource constrained system
DYMO has the capability to replace the existing AODV A lower value of this metric indicates energy efficient
routing from ZigBee stack. system
PERFORMANCE METRIC Total Charge Consumed
This metric specifies the total charge consumed in mAHr
We have chosen four performance metrics to analyze and during the simulation. This metric is computed with the
compare the performance of both protocols AODV and help of following equation
DYMO. These metrics are as follows:
Total Charge Consumed = ∑1 ( IBi − FBi )
N

Average Throughput at base station


where for ith node
It measures the average number of bytes received at the
base station for constant load at every node. It is calculated IBi = Initial Battery Capacity of node i in mAHr
using the following formula
FBi = Initial Battery Capacity of node i in mAHr
N

∑ B *8
i=2
i
bits This metric measures the current drainage of each node
Throughput at Base station = and will indicate the total battery consumed during session.
(TL − TF ) Sec
A lower value of this metric will indicate a power efficient
Where, system.

Bi = Number of bytes received from ith node SIMULATION MODEL AND PARAMETERS.
TL = Time last packet received by base station We have used QualNet 5.1 [17] for generating random
TF = Time first packet received by base station scenario of varying number of nodes with different seeds.
We varied the number of nodes starting from 50 numbers
A higher value of this metric will indicate higher number through 250 numbers in steps of 50. For each deployment
of data packets received by base station for the given we have generated 100 different set of spatial data for 100
simulation time. X 100m2 area. Position of the sink is assumed to be at the
center of the sensor field. Each node sends 80 packets of
Average end to end delay 512 bytes each to the base station. Hence at the application
layer we have used CBR where each node sends 80
It measures the average delay for packet transmission packets to base station at every a 10 sec interval. Total
between node and base station. It is calculated using the simulation time is kept to be 900sec. We have used IEEE
following formula 802.15.4 MAC and physical radio. The value of duty cycle
∑ (t − tsi )
P
ri
is varied by selecting different values of BO and SO as
Average end − to − end Delay = i =1
explained in section II. The value of SO is taken as 3, and
P value of BO is varied from 3 through 5 resulting duty cycle
where for each ith packet , of 100%, 50%, and 25% respectively. Since in our study
tri = time when base station received the packet we have considered three different duty cycles and two
different routing protocols, which results in six different
tsi = time when s ensor node sent the packet combinations of each. Each combination is given unique
P = total number of packets received name e.g. AODV with 100% duty cycle is termed it as
AODV 100% or DYMO with 25% duty cycle is termed as
A lower value of this metric will indicate low latency in
DYMO 25% and so on.
system .
Hence six different scenarios generate are summarized in
table II
Table II Average energy consumption
Scenario Duty Cycle Routing Protocol Scenario Name 1
1 100% AODV AODV 100% AODV 100%
2 50% AODV AODV 50% AODV 50%
0.9
3 25% AODV AODV 25% AODV 25%

Average Energy Consumed in (mJ)


4 100% DYMO DYMO 100% DYMO 100%
0.8
5 50% DYMO DYMO 50% DYMO 50%
6 25% DYMO DYMO 25% DYMO 25%
0.7

RESULTS AND DISCUSSION 0.6

0.5
Average Throughput at base station
0.4
3500

0.3
3000
0.2
50 100 150 200 250
2500 Number of Nodes
Throughput (bps)

DYMO with 25% duty cycle consumes the least amount of


2000
AODV 100% energy or we cans say that energy efficiency is best for
AODV 50%
DYMO 25%. Energy consumption increases with number
AODV 25%
DYMO 100% of node.
1500 Total Charge Consumed
DYMO 50%
DYNO 25% 70
AODV 100%
1000
AODV 50%
60 AODV 25%
500 DYMO 100%
50 100 150 200 250 DYMO 50%
50
Number of Nodes DYMO 25%

We plotted the value of average throughput given by 40

equation 1 with respect to number of nodes considered in


the scenario for six different case considered as shown in 30

table 1 From throughput point of view DYMO with 25%


duty cycle outperforms rest. Higher throughput obtained 20
for DYMO with 25% duty cycle is due to the fact that
when the node is in the sleep mode the packets are 10
buffered and hence the loss of packets is minimized
whereas in AODV no such mechanism exists. 0
50 100 150 200 250
Average end to end delay
1100
DYMO 100% Total charge consumption plot shows that DYMO 25% has
1000 DYMO 50% least current drainage from batteries where as AODV
DYMO 25%
AODV 100%
100% has maximum drainage.
Average End to end Delay (ms)

900
AODV 50% CONCLUSION
800 AODV 25%
Average throughput is highest for DYMO when it is used
700
with 25% duty cycle at MAC; also energy consumption is
600 least so we can say that most energy efficient routing of
data will occur when we use DYMO as routing protocol
500
with 25% duty cycle at MAC level. End to End delay is
400 more at low duty cycle DYMO so there exists a trade off
300
between delay and throughput. We can conclude that for
delay tolerant sensor networks DYMO with 25% duty
200 cycle presents more attractive option for energy efficiency
50 100 150 200 250
Number of Nodes as compare to present day AODV routing.
Average end to end delay, is minimum for DYMO with
100% duty cycle this is due to the fact that all the nodes are REFERENCES
active all the. For low latency application where energy [1] I. F. Akyildiz, W. Su, Y.Sankasubramaniam, and E. Cayirci , “
Wireless Sensor Networks: A Survey,” Computer Networks, 38 pp
efficiency may not be a requirement this can be a better 393–422, 2002.
choice. [2] C.Y. Chong and S.P.Kumar, “Sensor Networks: Evolution,
Opportunities and Challegenges,” Proceedings of IEEE, Vol. 91,
No8,pp 1247-1256, August 2003
[3] J. Yick, B. Mukherjee, and D. Ghosal, “Wireless Sensor Networks
Survey,” Computer Networks, 52 pp 2292–2330, 2008.
[4] Embedded WiSeNTs Consortium, Embedded WiSeNTsResearch
Roadmap (Deliverable 3.3). [Online]. Available:
http://www.embedded-wisents.org
[5] ON World Inc., “ Wireless Sensor Networks – Growing Markets
Accelating Demands,” March 2010 [Online] Available:
http://www.onworld.com/html/wirelesssensorsrprt2.htm
[6] R.Want, K.I. Farkas,and C. Narayanaswami, “Guest Editors'
Introduction: Energy Harvesting and Conservation,” IEEE
Pervassive Computing, Vol.4, No.1, pp 14-17, 2005.
[7] G.J Pottie and W.J. Kaiser, “Embedding the internet: Wireless
Integrated Network Sensors,” Communication of ACM, Vol.43, No.
5, pp 51-58, May 2000.
[8] IEEE 802 Working Group, Standard for Part 15.4: Wireless
Medium Access Control (MAC) and Physical Layer (PHY)
Specifications for Low-Rate Wireless Personal Area Networks (LR-
WPANs), ANSI/IEEE Std. 802.15.4. Sep.2006.
[9] ZigBee Alliance. [online]. Available : http://www.zigbee.org
[10] ZigBee Specification Verssion 2007, ZigBee Document
0534747r17, January 2008. [online]. Available :
http://www.zigbee.org
[11] C.E Perkins, E.M. Royer, “Ad-hoc on-demand distance vector
routing,” in proceeding of 2nd IEEE workshop on Mobile
Computing Systems and Applications (WMSA 1999), pp. 90-100,
feb 1999.
[12] I. Chakeres and C. Perkins, “Dynamic MANET On-demand
(DYMO) Routing,” IETF Internet Draft, V.19, March 2010.
[13] S.T. Sheu, Y.Y. Shih, and L.W. Chen, “An Adaptive Interleaving
Access Scheme (IAS) for IEEE 802.15.4 WPANs,”in the
proceedings of IEEE Vehicular Technology Conference
(VCT2005), Vol. 3, pp 1523-1527, 2005.
[14] Y. K. Huang, A.C.Pang, and H. N. Hung, “A Comprehensive
Analysis of Low Power Operation of Beacon-Enabled IEEE
802.15.4 Wireless Networks,” IEEE Transactions on Wireless
Communications, Vol. 8, No. 11, pp.5601-5611, November 2009.
[15] C.Perkins, E.B.Royer and S.Das,”AdHoc On-Demand Distance
Vector (AODV) Routing”, RFC 3561, IETF Network Working
Group, July 2003.
[16] T.G.Basavaraju and Subir Kumar Sarkar, “Adhoc Mobile
WirelessNetworks: Principles, Protocols and Applications”,
Auerbach Publications, 2008.
[17] “Qualnet 4.5 user’s Guide”, [online] Available
http://www.scalablenetworks.com/

Das könnte Ihnen auch gefallen