scispace - formally typeset

Proceedings ArticleDOI

XD: A Cross-Layer Designed Data Collection Mechanism for Mission-Critical WSNs in Urban Buildings

18 May 2009-pp 399-404

TL;DR: This work proposes a cross-layer designed data dissemination mechanism, referred to as Cross-Layer Diffusion (XD), in which notions in the path discovery (routing) component are exploited by data forwarding (MAC) component to improve the delivery rate and transmission delay.

AbstractAs the R&D experience accumulates, there is a rising interest of wireless sensor network (WSN) deployment in the urban environment. For mission critical applications such as healthcare and workplace safety, in particular, it is essential that the data dissemination mechanisms satisfy two important quality of service (QoS) requirements: (1) high delivery rate and (2) low transmission delay. Proposed in this work is a cross-layer designed data dissemination mechanism, referred to as Cross-Layer Diffusion (XD), in which notions in the path discovery (routing) component are exploited by data forwarding (MAC) component to improve the delivery rate and transmission delay. Using traces collected from a prototype WSN deployed in urban environment, we compare XD to the state-of-the-art mechanisms and find that XD is not only more efficient but also more practical.

Topics: Wireless sensor network (53%), Mission critical (53%), Transmission delay (52%), Dissemination (51%), Quality of service (51%)

Summary (2 min read)

Introduction

  • Such redundancy is desired by mission-critical applications in that the data delivery rate tends to be high and the end-to-end delay tends to be low.
  • The conventional TDMA mechanisms do not scale to the density of the WSNs.
  • In their design, nodes of the same magnetic charge send data at the same time slot.
  • The authors describe the of XD in detail and provide the rationale o layer designed mechanism achieves relia sensor data collection.

A. Path Discovery

  • Consider th magnet and the data as metallic nails.
  • T attracted towards the sink according to the m as the nails are attracted towards the magn in MD llision avoidance mechanism is llected from an the simulations, tacks, MD with delivery rate and use of the multif-the-art TDMA commonly-used indicate that XD Zigbee stack.
  • Ll to the state-ofing the TDMA nces support that ction mechanism nd yet effective.
  • The mag in the interest broadcast phase su disseminated towards the sink in the At the interest broadcast phase, th charge is set the highest charge an packet to its neighbors.
  • The node then records this mag the interest message to its neigh magnetic charges from the sink to s data in the reverse direction, from h points in the magnetic field.

B. Data Forwarding

  • XD incorporates a hybrid mechanism that utilizes informati Even if there is no contention in delivery rate may still suffer from commonly seen in wireless network proper magnetic charges etic influence of the data determined by the hop propagated based on the high-charge nodes.
  • In Figure 2, node A and B are sending nodes and the packets collide at node C. The number on the node indicates the assigned magnetic charge.
  • A significant amount of the collisions can be further reduced by separating the sending time of the nodes at different levels, hinting the benefit of a TDMA strategy to the design.
  • Packets travelling multiple disjoint paths may collide at the sink node.
  • The data delivery rate at the low traffic load case is effectively raised to 100% while the end-to-end delay is slightly compromised.

B. MD with ZMAC:

  • ZMAC is a hybrid CSMA and TDMA mechanism.
  • The schedule used by its TDMA component is derived by the DRAND [9] algorithm which requires topological information in the two-hop neighborhood.
  • In ZMAC, although every node has its slots for transmission, it allows nodes competing sending data at the slots which are not assigned.
  • From mid-load cases and on, the end-to-end delay is raised to the scale of seconds and the data delivery rate begins to drop afterwards.
  • Combined with the multiple shortest paths discovered by MD, XD performs, in terms of data delivery rate and end-toend delay, better than the Zigbee stack and just as well to the state of the art.

Did you find this useful? Give us your feedback

...read more

Content maybe subject to copyright    Report

XD: A Cross-Layer Designed Data Collection Mechanism for Mission-Critical
WSNs in Urban Buildings
Chieh-Ting Huang
a
, Tsung-Han Lin
b
, Ling-Jyh Chen
c
, Polly Huang
a
a
Department of Electrical Engineering
National Taiwan University
b
School of Engineering and Applied Sciences
Harvard University
c
Institute of Information Science
Academia Sinica
r96921024@ntu.edu.tw, thlin@eecs.harvard.edu, cclljj@iis.sinica.edu.tw, phuang@cc.ee.ntu.edu.tw
Abstract—As the R&D experience accumulates, there is a
rising interest of wireless sensor network (WSN) deployment in
the urban environment. For mission critical applications such
as healthcare and workplace safety, in particular, it is essential
that the data dissemination mechanisms satisfy two important
quality of service (QoS) requirements: (1) high delivery rate
and (2) low transmission delay. Proposed in this work is a
cross-layer designed data dissemination mechanism, referred
to as Cross-Layer Diffusion (XD), in which notions in the path
discovery (routing) component are exploited by data
forwarding (MAC) component to improve the delivery rate
and transmission delay. Using traces collected from a
prototype WSN deployed in urban environment, we compare
XD to the state-of-the-art mechanisms and find that XD is not
only more efficient but also more practical.
Keywords- Urban Wireless Sensor Network; Data
Dissemination, Cross-Layer Design
I. INTRODUCTION
Envisioning a new generation of mission-critical
applications in the urban environment, we seek mechanisms
that provide reliable and timely transmissions of sensor data.
The two QoS requirements, (1) high data delivery rate and (2)
low end-to-end transmission delay, need to be jointly
considered when devising such mechanisms. It is, however,
not trivial to satisfy the two requirements at the same time.
For high delivery rate, one often takes the ARQ approach [1]
to retransmit lost data. This adds to the end-to-end delay and,
in a way, trades off the other QoS requirement for mission-
critical applications. Alternatively, one may take the FEC
approach [2] which transmits redundant data to raise the
probability of having at least one copy arriving at the data
sink. This approach appears to allow timely delivery of data
in the presence of loss, but it could also come back to haunt
the network in terms of both loss and delay if the degree of
redundancy is not cautiously administered. Taking an FEC-
with-care approach, we propose a cross-layer designed data
collection mechanism, referred to as Cross-Layer Diffusion
(XD).
XD is a mission-critical data collection mechanism with
(1) opportunistic redundancy and (2) collision avoidance.
The path discovery part of XD is based on a multi-path
shortest path routing mechanism referred to as Magnetic
Diffusion (MD) [3]. In that, the network establishes a
magnetic field across the WSN, with decreasing magnetic
charges radiating from the sink. Sensor data, in turn, travel
the network tending to the nodes with higher magnetic
charges until reaching the sink, mimicking how metallic
objects (data) are attracted by a magnet (sink). If there exist
multiple shortest paths in the WSN topology, all of such
paths will be discovered by the mechanism. As a result,
sensor data are disseminated with redundancy in a way that
duplicates will travel multiple shortest paths. Such
redundancy is desired by mission-critical applications in that
the data delivery rate tends to be high and the end-to-end
delay tends to be low. Because the degree of redundancy
depends on the network topology, we refer to this property of
XD as opportunistic redundancy.
The data forwarding part of XD is co-designed with the
path discovery component. In that, a hybrid TDMA and
CSMA mechanism is proposed to administer the forwarding
of sensor data. The rationale behind such a design is that,
with the opportunistic redundancy, there will be a higher
amount of bits travelling the WSN which increases the
chance of collision. TDMA-based solutions are effective
avoiding collisions in high traffic load cases. However,
conventional TDMA divides the time slots in a cycle by
individual nodes in the neighborhood. This results in higher
end-to-end delays in densely-deployed WSNs. The
conventional TDMA mechanisms do not scale to the density
of the WSNs.
Our contribution lies in the design of a novel TDMA
mechanism utilizing the magnetic charges established by the
path discovery component. In our design, nodes of the same
magnetic charge send data at the same time slot. A cycle is
divided into three time slots only. A node with magnetic
charge C transmits data at time slot C mod 3. This prevents
data coming from a level up or a level down to collide at the
middle level (hidden terminal). Furthermore, to avoid data
collisions among the nodes of the same charge, a CSMA
mechanism is applied within each level. I.e., each node
listens to the channel and makes sure other nodes at the same
charge level are indeed idle before transmitting. The TDMA

Figure 1. Flow of Control and Data Packets
and CSMA combined contribute to the co
property of XD.
This simple, cross-layer designed
implemented in ns-2 [4]. Using traces c
o
WSN deployed in an urban building to driv
e
we compare XD to two other protocol
s
ZMAC [8] and Zigbee [5] in terms of data
end-to-end delay. The former represents the
path discovery mechanism with a state-
o
MAC protocol and the latter represents th
e
industrial standard. The simulation results
performs significantly better than the
Additionally, XD performs comparably we
the-art while the complexity of establis
h
schedule is significantly lower. These evid
e
XD is an effective and
p
ractical data colle
for mission-critical WSNs.
II. M
ECHANISM
Cross-layer Diffusion (XD) is simple
a
The path discovery component impleme
shortest path routing mechanism. The
component implements a hybrid CS
M
mechanism based on the notion in the
component. In this section, we describe th
e
of XD in detail and provide the rationale
o
layer designed mechanism achieves reli
a
sensor data collection.
A. Path Discovery
The design of MD is inspired by the
magnets and metallic objects. Consider t
h
magnet and the data as metallic nails.
T
attracted towards the sink according to the
m
as the nails are attracted towards the mag
n
in MD
llision avoidance
mechanism is
o
llected from an
e
the simulations,
s
tacks, MD with
delivery rate and
use of the multi-
o
f-the-art TDMA
e
commonly-used
indicate that XD
Zigbee stack.
ll to the state-of-
h
ing the TDMA
e
nces support that
ction mechanism
a
nd yet effective.
n
ts a multi-path
data forwarding
M
A and TDMA
path discovery
e
two components
o
f how the cross-
a
ble and timely
physics between
h
e data sink as a
T
he data will be
m
agnetic field just
n
et. The magnetic
field is established by setting up th
e
on the sensor nodes within the mag
n
sink. The strength of the charge i
s
distance to the sink. The data will
be
magnetic field from low-charge t
o
this principle, data traveling towar
d
the sensor nodes forward data comi
n
with smaller charges. Forwarding
selected by MD are the shortest in
there can be multiple next hops t
o
MD sends data in an optimal multi-
p
MD operates in two phases: t
h
data propagation. Figure 1 shows
data packets in MD. The number o
n
magnetic charge assigned. The ma
g
in the interest broadcast phase s
u
disseminated towards the sink in th
e
At the interest broadcast phase, t
h
charge is set the highest charge a
n
packet to its neighbors. When a
n
message, it decreases the magnetic
one. The node then records this ma
g
the interest message to its neig
h
magnetic charges from the sink to
s
data in the reverse direction, from
h
points in the magnetic field.
The second phase, data dissemi
n
data and broadcasts packets wit
h
Therefore, the receiving node can
carried in the data where the d
a
whether to forward the data furth
e
node compares the magnetic cha
r
charge of itself. If the magnetic c
h
the node increases the magnetic c
h
and continues to disseminate the da
t
node receives a duplicate data or
t
charge is smaller than the node’s, th
B. Data Forwarding
XD incorporates a hybrid
mechanism that utilizes informat
i
Even if there is no contention i
n
delivery rate may still suffer fro
m
commonly seen in wireless networ
k
Figure 2. Types of Collision in MD
e
proper magnetic charges
n
etic i
n
fluence of the data
s
determined by the hop
e
propagated based on the
o
high-charge nodes. By
d
s the center of attraction,
n
g strictly from the nodes
data this way, the paths
hop count. Furthermore,
o
forward the data. Thus,
p
ath fashion.
h
e interest broadcast and
the flow of interest and
n
each node indicates the
g
netic field is established
u
ch that the data can be
e
data propagation phase.
h
e sink whose magnetic
n
d broadcasts the interest
n
ode receives an interest
charge in the interest by
g
netic charge and forward
h
bors. The decrementing
s
ource will guide flow of
h
igh-charge to low-charge
n
atio
n
, a node senses the
h
the charge of itself.
identify from the charge
a
ta is coming from and
e
r down the stream. The
r
ge of the data with the
h
arge in the data smaller,
h
arge of the data by one
t
a. On the other hand, if a
t
he data whose magnetic
e data will be discarded.
TDMA-CSMA MAC
i
on established b
y
MD.
n
the network,
t
he data
m
losses due to bit errors
k
s. Our design objective

Fi
g
ure 5. Hidden Terminal Effect around Sink.
is thus not to establish a 100% contention-free network but to
avoid as many collisions as possible with redundancy. More
specifically, the TDMA part of the design eliminates
collisions from hidden terminals two magnetic charge levels
apart and its CSMA counterpart avoids collisions between
neighboring nodes.
To elaborate in more detail the design rationale, we
classify collisions in MD networks into six canonical types.
In Figure 2, node A and B are sending nodes and the packets
collide at node C. The number on the node indicates the
assigned magnetic charge. Type I, II and III are cases where
sending nodes are at the same level. In that, the sending
nodes can be one level higher, at the same level, or one level
lower than the receiving node. Type I and II collisions are
not critical because these data are traveling against the
magnetic field. Type III collisions are, however, problematic.
These collisions can be reduced partly by applying CSMA at
nodes within the same level since nodes at the same level
tend to be nearby. Note that, though, there remains a slight
chance where nodes at the same level are away from each
other’s carrier sense range.
Type IV and V are cases where sending nodes are of
adjacent magnetic charges. Again, Type IV is not critical
given the transmission is not effective for data collection at
the sink. Type V collisions, the critical ones, can be reduced
by CSMA to a certain degree because nodes with adjacent
magnetic charges can hear each other relatively easily. A
significant amount of the collisions can be further reduced by
separating the sending time of the nodes at different levels,
hinting the benefit of a TDMA strategy to the design. In
Type VI, where the two sending nodes are two levels apart,
nodes of higher and smaller level become hidden terminals
to each other. These collisions may not be eliminated by
carrier sense. The sending time of nodes two levels apart
needs to be separated to avoid collisions effectively. To
avoid type V and VI collisions effectively, TDMA based on
the magnetic charge levels is applied.
XD uses three time slots for the TDMA schedule. The
slot assignment is based on each node’s magnetic charge. In
the interest broadcast phase of XD, the interest messages,
originated from the sink, carry magnetic charges
decremented at each hop. When a node receives the interest
message, the node records the magnetic charge in the
message and the TDMA slot for the node is assigned to n
mod 3, where n is the recorded magnetic charge.
An example of the TDMA mechanism is shown in Figure
3. The numbers in the circles are the magnetic charges after
the interest broadcast phase. In slot two, only nodes with
charge 8 and 5 can send packets and the packets can be
received by the nodes that are downstream, e.g. nodes with
charge 9 and 6 respectively. At this time, nodes with charge
10 and 7 will go into the sleep mode since receiving packets
from their downstream neighbors is not necessary. In the
next time slot, those nodes, who receive packets from
upstream nodes at previous slot, will go into the transmit
mode and transmit the packets to the nodes at the next level.
Furthermore, XD may discover multiple paths to deliver
data. Packets travelling multiple disjoint paths may collide at
the sink node. Consider an urban WSN shown in Figure 5.
The sender sends packets to the sink via two disjoint paths.
Though node A, B and C are neighbors of the sink, node C is
separated from node A and B by a corner in an indoor
environment. In this case, A and B will not send data
simultaneously due to CSMA. However, node B and C
cannot hear each other thus collisions may still occur when
node C also sends packets. In order to prevent data collisions
at the sink node, XD further refines the TDMA schedule at
the sink node. The sink node’s neighbors are grouped. Each
group shares the same carrier sense region. By collecting
neighbors’ information at the sink, the sink can determine
how many groups and the appropriate schedule separating
the sending of data from each group. Each time slot is
divided into multiple sub-slots depending on the number of
groups there are. An example of the refined schedule is
shown in Figure 4. The nodes with charge 9 are the sink
neighbors. That slot is further divided into two for two
groups of neighbors, i.e., 9a and 9b. When 1a transmits
packets, 9b goes to sleep. When 9a goes to sleep, 9b
transmits packets.
III. T
RACE-DRIVEN SIMULATION
To examine the performance of XD, we conduct a set of
trace-driven simulations using ns-2 [4]. This methodology
enables us to (1) repeat the tests, (2) reuse the existing
simulation implementation, and in the meantime (3) capture
the effect of packet losses occurring in realistic urban
buildings.
N
odes with Magnetic Charge
10
9a
8 7 6 5
9b
T RS R S T
At slot 2
R
R ST S T R
At slot 1
S
S
R
T R S
At slot 0
T
R
S
R S T
At slot 2
S
T
R
T:Transmit
R:Receive
S:Sleep
Figure 4. Extended TDMA Schedule
R
S
T
S T R
At slot 1
S
T
R T R S
At slot 0
S
S
N
odes with Magnetic Charge
10 9 8 7 6 5
R S T
S T R
At slot 1
S T R
T R S
At slot 0
T R S
R S T
At slot 2
T:Transmit
R:Receive
S:Sleep
Figure 3. Three-Level TDMA Schedule

Figure 6. Testbed Topology.
Figu
r
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Reception Rate (%)
40.96
51.20
68.27
78.77
85.33
93.09
102.40
113.78
first-hop node
Recv
Recv&Colli
Colli
None
Our implementation of XD includes (1
)
hybrid MAC. The magnetic charges s
component are communicated to the MA
C
the magnetic-level-
b
ased TDMA sched
u
compared to two suites to data collection
p
the well-known Zigbee stack. The MA
C
Zigbee is a simplified version of the nati
v
module [6] in ns-2, and the routing comp
from the AODV module [7] in ns-2. The o
t
of MD and a state-of-the-art MAC, ZMA
C
feature in our simulation approach is the
u
p
atterns collected from an actual WSN. In
s
in wireless link propagation models imple
m
implemented a trace-driven physical laye
allows us to simulate realistic wireless p
a
collisions in an indoor environment.
Figure 6 illustrates the actual WSN u
s
traces for the physical layer module. The te
s
on the 6
th
floor of the Electrical Engine
e
building of NTU.. There are 12 nodes in
measure the packet reception rate (PRR)
o
programming 11 nodes as receiver and 1
Each of the 12 nodes will take turn to b
e
PRR is obtained by calculating the ratio
o
p
ackets received to the number of pac
k
simulation, when a node sends a pac
k
determines based on the measured PRR w
h
can be heard at the intended receiver
.
p
reliminary measurement results, we obser
v
wireless links are often asymmetric, very
d
propagation models, which are more app
r
space transmissions, implemented in ns-2.
Figure 7. CDF of End-to-End Dela
y
10
-3
10
-2
10
-1
10
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Delay (s)
Reception Rate (%)
Data Rate (Kbps)
r
e 8. Distribution of Packets Observed at Each Node.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
40.96
51.20
68.27
78.77
85.33
93.09
102.40
113.78
second-hop node
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
40.96
51.20
68.27
78.77
sink
)
MD and (2) the
s
et by the MD
C
component for
u
ling. XD is
p
rotocols. One is
C
component of
v
e IEEE 802.15.4
onent is adopted
t
her suite consists
C
[8]. One unique
u
se of packet loss
s
tead of the buil
t
-
m
ented in ns-2, we
e
r module which
acket losses and
s
ed to collect the
s
tbed is deployed
e
ring Department
the testbed. We
of each node by
node as sender.
e
the sender. The
o
f the number of
k
ets sent. In the
k
et, other nodes
h
ethe
r
the packet
.
Based on the
v
e that the indoor
d
istinctive of the
r
opriate for open
The exact topology of the testbe
d
simulations as well. The bottom le
f
data sink. The data sink’s charge i
s
charge after the interest broadcas
t
node is shown as the number in
t
nodes of the lowest charge, 7. The
o
source. In each simulation, the so
u
before the simulation terminates.
IV. P
ERFORMANC
E
We first analyze the effect of i
mechanism around the sink node a
n
based on the magnetic charges. Fi
g
end-to-end delay in MD with sim
p
simulations, we control all factors
b
source node, from 22.76kbps to
observe that the delay increases an
d
as the source data rate increases. In
rate is below 68.27kbps, the end-
t
low, but (2) the data delivery rate
best. This is to our surprise kno
w
IEEE 802.15.4 is 250kbps at
t
utilization of the 22.76kbps worklo
a
(two charge-8 nodes help forwardin
g
To understand the reason of suc
h
when the workload is low, we clas
s
source into four types by how they
along the transmission path. Beca
u
travel multiple paths, it is often a n
o
individual packets. If all dupli
c
successfully received at the node, t
h
y
in MD with CSMA.
0
10
1
10
2
22.76
25.60
40.96
68.27
85.33
93.09
113.78
85.33
93.09
102.40
113.78
node
d
is used to drive the ns-2
f
t node in Figure 6 is the
s
set to 10. The magnetic
t
phase of MD for each
t
he node. There are two
o
ne on the lef
t
is the data
u
rce sends 1000 packets
E
ANALYSIS
m
plementing the TDMA
n
d the TDMA mechanism
g
ure 7 shows the CDF of
p
le CSMA. In this set of
b
ut the sending rate of the
113.78kbps. We can
d
reception rate decreases
particular, (1) when data
t
o-end delay is generally
is only about 80-90% at
w
ing that the capacity of
h
e physical level. The
a
d case is only about 20%
g
).
h
a low data deliver
y
rate
s
if
y
packets sent from the
a
re received a
t
each node
u
se XD allows packets to
o
de receives duplicates of
c
ates of a packet are
h
e packet is categorized as

Figure 9. CDF of End-to-End Delay in MD with TDMA around Sink.
10
-3
10
-2
10
-1
10
0
10
1
10
2
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Delay (s)
Reception Rate (%)
22.76
25.60
40.96
68.27
85.33
93.09
113.78
Figure 10. CDF of End-to-End Delay in XD
10
-3
10
-2
10
-1
10
0
10
1
10
2
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Delay (s)
Reception Rate (%)
22.76
25.60
40.96
68.27
85.33
93.09
113.78
the first type (Recv). If a certain duplicate of a packet arrives
while at least one of the duplicates is collided, the packet is
categorized into the second type (Recv & Colli). The third
type is for packets whose duplicates are all collided (Colli).
The last type is for packets that are never sent by the
upstream nodes (None). These are the packets that might
have been collided at upstream nodes. The sum of the four
types is 100% which is the total number of packets sent by
the data source.
The three plots in Figure 8 depict the portion of the
packets of each type received at nodes from the first-hop
neighbor of the source towards the sink. Each column
represents the portion of the packets of each type at different
sending rate. One can observe that, at the first-hop node,
there are an increasing number of packets being collided as
the workload increases. Most of the packets are collided due
to the hidden terminal effect between the source and the
second-hop nodes. The portion of packets that are never seen
at the first hop node also increases as the sending queue at
the source node builds up faster and the packets are
eventually dropped at the source node before they get to
travel the network at all. The packets collided at the first hop
and the packets never sent from the source will not reach the
second-hop node at all. The amount is indicated as the type
four packets (None) in the middle plot. There is a minor
amount of collisions at the second-hop nodes which results
in small portion of type two and type three packets. This is
because the sink node does not forward data further.
Therefore, there is no observation of collisions due to hidden
terminals between the sink and the first-hop node at the
second-hop node.
The behavior at the sink node, i.e., the third-hop node, is
particularly interesting. One would expect the portion of type
four packets to be at least the amount observed in the
second-hop node. The simulation results reflect the property
of multi-path routing where packets might travel through
other paths to reach the data sink. In this case, a significant
amount of data eventually reaches the data sink. Also due to
the fact that packets may travel multiple paths in the network,
there is a slightly higher amount of collisions observed at the
sink node. These contribute to the higher amounts of type
two and type three packets observed at the sink node.
The performance of multi-path routing with simple
CSMA is rather interesting in that (1) the multiple paths
result in a higher delivery rate at the sink when the workload
is high but in the meantime (2) the hidden terminal effect
around the sink node also results in collisions which lowers
the data delivery rate when the workload is low. We then
turn on the TDMA around the sink node to eliminate the
hidden terminal effects observed above. Figure 9 shows the
CDF of end-to-end delay in MD with CSMA and TDMA
around sink. The data delivery rate at the low traffic load
case is effectively raised to 100% while the end-to-end delay
is slightly compromised.
Figure 10 shows the CDF of end-to-end delay of
complete XD. Because XD schedules both sink neighbors
and nodes from different magnetic charges, it eliminates both
hidden terminal problem at the sink node and the
intermediate nodes (the first-hop and second-hop nodes).
Important performance properties observed are as follows. (1)
The data delivery rate in low-load cases are almost 100%. (2)
In mid-load cases, e.g., 68.27kbps and 85.33kbps, XD helps
raising the data delivery rate. This, however, trades off the
end-to-end delay, a problem general to TDMA-based
solutions. Packets need to wait in the queue for the time slots
assigned and such waiting repeats at each node along the
transmission path. The queuing delay accumulates. (3) When
the data rate is high, i.e., the traffic is close to the network
capacity, the queue may overflow and result in packet drops,
a problem that can only be alleviated by extending the
network capacity.
V.
COMPARISON
We next compare XD to two other protocol stacks for
sensor data collection. The first stack is Zigbee, using
AODV for path discovery and IEEE802.15.4 for data
forwarding. The other stack uses MD for path discovery and
ZMAC for data forwarding.
A. AODV with IEEE802.15.4:
AODV is an on-demand single-path routing protocol.
When a node needs to transmit data, AODV lookups its
routing table to see is there exists a routing path to the
destination. If not, the packet is buffered at the AODV queue
(not the queue at MAC layer) while the mechanism begins to
discover a routing path. This introduces a certain amount of
delay.

Citations
More filters

01 Jan 2005

51 citations


Proceedings ArticleDOI
Yu Gu1, Tian He1
21 Jun 2010
TL;DR: This paper introduces novel solutions for bounding sink-to-node communications in energy-harvesting sensor networks and theoretically proves its NP-Hardness and in approximability property, followed by an efficient heuristic solution.
Abstract: In energy-harvesting sensor networks, limited ambient energy from environment necessitates sensor nodes to operate at a low-duty-cycle, i.e., they communicate briefly and stay asleep most of time. Such low-duty-cycle operation leads to orders of magnitude longer communication delays in comparison with traditional always-active networks, imposing a new challenge in many time-sensitive sensor network applications (e.g., tracking and alert). In this paper, we introduce novel solutions for bounding sink-to-node communications in energy-harvesting sensor networks. We first present an optimal solution for the sink-to-one case and its distributed implementation. For the sink-to-many case, we theoretically prove its NP-Hardness and in approximability property, followed by an efficient heuristic solution. We have evaluated our design with both extensive simulation and a TinyOS/Mote based implementation. Compared with an improved version of a state-of-the-art design, our delay maintenance design effectively provides E2E delay guarantees while consuming as much as 60% less energy.

48 citations


Cites background from "XD: A Cross-Layer Designed Data Col..."

  • ...For many of those operations, there is usually also a delay bound associated with them and require the messages sent out by the sink node to be received at destined receivers within a designated time bound [14]–[16]....

    [...]


Journal ArticleDOI
TL;DR: This work investigates a fundamental scheduling problem of both theoretical and practical importance, called multi-task schedulability problem, to determine the maximum number of tasks that can be scheduled within their deadlines and work out such a schedule.
Abstract: In many sensor network applications, multiple data forwarding tasks usually exist with different source-destination node pairs. Due to limitations of the duty-cycling operation and interference, however, not all tasks can be guaranteed to be scheduled within their required delay constraints. We investigate a fundamental scheduling problem of both theoretical and practical importance, called multi-task schedulability problem, i.e., given multiple data forwarding tasks, to determine the maximum number of tasks that can be scheduled within their deadlines and work out such a schedule. We formulate the multi-task schedulability problem, prove its NP-Hardness, and propose an approximate algorithm with analysis on the performance bound and complicity. We further extend the proposed algorithm by explicitly altering duty cycles of certain sensor nodes so as to fully support applications with stringent delay requirements to accomplish all tasks. We then design a practical scheduling protocol based on proposed algorithms. We conduct extensive trace-driven simulations to validate the effectiveness and efficiency of our approach with various settings.

7 citations


Cites background from "XD: A Cross-Layer Designed Data Col..."

  • ...We investigate a fundamental scheduling problem of both theoretical and practical importance, called multitask schedulability problem, i.e., given multiple data forwarding tasks, to determine the maximum number of tasks that can be scheduled within their deadlines and work out such a schedule....

    [...]


Journal ArticleDOI
TL;DR: This paper proposes autility- based delay bounded scheme for data forwarding MaxOpUtility-based scheme, which offers the ability to increase reliability through relay nodes selection as well as to ensure the timeliness of messages sent within a designated time bound.
Abstract: In many sensor network applications, sink node needs to actively communicate with other sensor nodes in order to perform data forwarding operations. For those applications, there is usually adelay-bounded associated with them and require the messages sent to be received within a designated time bound. In energy harvesting sensor networks, limited energy from environment necessitates sensor nodes to operate at alow-duty-cycle. Sensor nodes work active briefly and stay asleep most of time. Such low-duty-cycle operation leads to communication delays in comparison with the always-active networks. In this paper, we address the data forwarding problems in an energy harvesting sensor network where energy efficiency and data freshness need to be balanced. To solve this problem, we propose autility-based delay bounded scheme for data forwarding MaxOpUtility-based scheme. MaxOpUtility scheme offers the ability to increase reliability through relay nodes selection as well as to ensure the timeliness. In add...

2 citations


Proceedings ArticleDOI
02 Dec 2013
TL;DR: The Centralized Cluster-based Location Finding (CCLF) algorithm is proposed to reduce the high latency in low-duty-cycle WSNs by finding a suitable position for the sink by requiring less operation time compared with the optimal algorithm.
Abstract: Low-duty-cycle mechanisms can reduce the energy consumption in wireless sensor networks. Many related researches have been made in recent years. In the low-duty-cycle environment, the latency of sending packets from a sink to each node is much longer than traditional WSNs because nodes stay asleep most of the time. In this paper, the Centralized Cluster-based Location Finding (CCLF) algorithm is proposed to reduce the high latency in low-duty-cycle WSNs by finding a suitable position for the sink. The algorithm mainly consisted of three steps: (1) Cluster construction, (2) The fast look-up table (FLU-table) construction, and (3) Sink location decision. The simulation results show that the performance of the CCLF algorithm approaches the performance of the optimal algorithm. Moreover, the CCLF algorithm requires less operation time compared with the optimal algorithm.

1 citations


References
More filters

Proceedings ArticleDOI
25 Oct 1998
TL;DR: The results of a derailed packet-levelsimulationcomparing fourmulti-hopwirelessad hoc networkroutingprotocols, which cover a range of designchoices: DSDV,TORA, DSR and AODV are presented.
Abstract: An ad hoc networkis a collwtion of wirelessmobilenodes dynamically forminga temporarynetworkwithouttheuseof anyexistingnetworkirrfrastructureor centralizedadministration.Dueto the limitedtransmissionrange of ~vlrelessnenvorkinterfaces,multiplenetwork“hops”maybe neededfor onenodeto exchangedata ivithanotheracrox thenetwork.Inrecentyears, a ttiery of nelvroutingprotocols~geted specificallyat this environment havebeen developed.but little pcrfomrartwinformationon mch protocol and no ralistic performancecomparisonbehvwrrthem ISavailable. ~Is paper presentsthe results of a derailedpacket-levelsimulationcomparing fourmulti-hopwirelessad hoc networkroutingprotocolsthatcovera range of designchoices: DSDV,TORA, DSR and AODV. \Vehave extended the /~r-2networksimulatorto accuratelymodelthe MACandphysical-layer behaviorof the IEEE 802.1I wirelessLANstandard,includinga realistic wtrelesstransmissionchannelmodel, and present the resultsof simulations of net(vorksof 50 mobilenodes.

5,118 citations


"XD: A Cross-Layer Designed Data Col..." refers background in this paper

  • ...About half of the packets are collided which results in a significant data delivery rate for mid-load and high-load cases....

    [...]


Book
01 Jan 1983
TL;DR: This book explains coding for Reliable Digital Transmission and Storage using Trellis-Based Soft-Decision Decoding Algorithms for Linear Block Codes and Convolutional Codes, and some of the techniques used in this work.
Abstract: 1. Coding for Reliable Digital Transmission and Storage. 2. Introduction to Algebra. 3. Linear Block Codes. 4. Important Linear Block Codes. 5. Cyclic Codes. 6. Binary BCH Codes. 7. Nonbinary BCH Codes, Reed-Solomon Codes, and Decoding Algorithms. 8. Majority-Logic Decodable Codes. 9. Trellises for Linear Block Codes. 10. Reliability-Based Soft-Decision Decoding Algorithms for Linear Block Codes. 11. Convolutional Codes. 12. Trellis-Based Decoding Algorithms for Convolutional Codes. 13. Sequential and Threshold Decoding of Convolutional Codes. 14. Trellis-Based Soft-Decision Algorithms for Linear Block Codes. 15. Concatenated Coding, Code Decomposition ad Multistage Decoding. 16. Turbo Coding. 17. Low Density Parity Check Codes. 18. Trellis Coded Modulation. 19. Block Coded Modulation. 20. Burst-Error-Correcting Codes. 21. Automatic-Repeat-Request Strategies.

3,814 citations


Book
01 Feb 1995
TL;DR: This work has shown that polynomials over Galois Fields, particularly the Hadamard, Quadratic Residue, and Golay Codes, are good candidates for Error Control Coding for Digital Communication Systems.
Abstract: 1. Error Control Coding for Digital Communication Systems. 2. Galois Fields. 3. Polynomials over Galois Fields. 4. Linear Block Codes. 5. Cyclic Codes. 6. Hadamard, Quadratic Residue, and Golay Codes. 7. Reed-Muller Codes 8. BCH and Reed-Solomon Codes. 9. Decoding BCH and Reed-Solomon Codes. 10. The Analysis of the Performance of Block Codes. 11. Convolutional Codes. 12. The Viterbi Decoding Algorithm. 13. The Sequential Decoding Algorithms. 14. Trellis Coded Modulation. 15. Error Control for Channels with Feedback. 16. Applications. Appendices: A. Binary Primitive Polynomials. B. Add-on Tables and Vector Space Representations for GF(8) Through GF(1024). C. Cyclotronic Cosets Modulo 2m-1. D. Minimal Polynomials for Elements in GF (2m). E. Generator Polynomials of Binary BCH Codes of Lengths Through 511. Bibliography.

1,930 citations


Journal ArticleDOI
Abstract: Network researchers must test Internet protocols under varied conditions to determine whether they are robust and reliable. The paper discusses the Virtual Inter Network Testbed (VINT) project which has enhanced its network simulator and related software to provide several practical innovations that broaden the conditions under which researchers can evaluate network protocols.

777 citations


Journal ArticleDOI
TL;DR: A hybrid MAC protocol for wireless sensor networks that combines the strengths of TDMA and CSMA while offsetting their weaknesses, ZMAC, which achieves high channel utilization and low latency under low contention and reduces collision among two-hop neighbors at a low cost.
Abstract: This paper presents the design, implementation and performance evaluation of a hybrid MAC protocol, called Z-MAC, for wireless sensor networks that combines the strengths of TDMA and CSMA while offsetting their weaknesses. Like CSMA, Z-MAC achieves high channel utilization and low latency under low contention and like TDMA, achieves high channel utilization under high contention and reduces collision among two-hop neighbors at a low cost. A distinctive feature of Z-MAC is that its performance is robust to synchronization errors, slot assignment failures, and time-varying channel conditions; in the worst case, its performance always falls back to that of CSMA. Z-MAC is implemented in TinyOS.

739 citations


"XD: A Cross-Layer Designed Data Col..." refers background or methods in this paper

  • ...One is C component of ve IEEE 802.15.4 onent is adopted ther suite consists C [ 8 ]....

    [...]

  • ...Using traces co WSN deployed in an urban building to drive we compare XD to two other protocol s ZMAC [ 8 ] and Zigbee [5] in terms of data end-to-end delay....

    [...]