09-10-2014, 09:43 AM
Adaptive Forwarding Delay Control for VANET
Data Aggregation
Adaptive.pdf (Size: 716.51 KB / Downloads: 8)
Abstract
—In-network data aggregation is a useful technique to reduce redundant data and to improve communication efficiency.
Traditional data aggregation schemes for wireless sensor networks usually rely on a fixed routing structure to ensure data can be
aggregated at certain sensor nodes. However, they cannot be applied in highly mobile vehicular environments. In this paper, we
propose an adaptive forwarding delay control scheme, namely Catch-Up, which dynamically changes the forwarding speed of nearby
reports so that they have a better chance to meet each other and be aggregated together. The Catch-Up scheme is designed based on
a distributed learning algorithm. Each vehicle learns from local observations and chooses a delay based on learning results. The
simulation results demonstrate that our scheme can efficiently reduce the number of redundant reports and achieve a good trade-off
between delay and communication overhead.
INTRODUCTION
VEHICULAR Ad hoc Networks (VANETs) have been
regarded as an emerging and promising field in both
industry and academia. It has potential to improve the
efficiency and safety of future highway systems. One good
example is traffic congestion detection. Once a vehicle
detects the number of its neighboring vehicles exceeds a
certain limit, it will broadcast a warning to the vehicles
following behind. This warning could travel a rather long
distance so that vehicles, possibly several kilometers away,
can have enough time to choose an alternate route. Since each
vehicle in VANETs is able to detect traffic conditions and
generate traffic reports distributedly and independently,
redundant reports can be generated and forwarded in the
network, consuming a considerable amount of bandwidth.
Bandwidth utilization is an important issue for VANET
communications. For example, in Dedicated Short Range
Communications [1] (DSRC), the communication range can
be up to 1,000 m, while the data rate is only 6 to 27 Mbps.
Imagine the rush hour traffic, it is very likely that there are
more than a hundred vehicles within the range of 1,000 m,
all of which share a quite limited bandwidth.
Data aggregation is a potential approach to improving
communication efficiency. It consists of a variety of adaptive
methods which can merge information from various data
sources into a set of organized and refined information. The
process of data aggregation can be performed in-network so
that communication overhead can be effectively reduced
soon after redundant reports are generated.
SYSTEM MODEL
In this section, we introduce the system model which is
applicable to our aggregation scheme.
Basic assumptions. We assume each vehicle is equipped
with an on-board computing device, a wireless radio, a GPS
device, and digital maps. Time in different vehicles can be
synchronized by GPS technologies. For the purpose of
neighbor discovery, each vehicle periodically broadcasts
beacons so that neighboring vehicles are aware of its
identity, speed, and GPS location.
Applications. Our work focuses on providing generic data
aggregation service for various applications, such as traffic
congestion detection, road condition (i.e., icy road surface)
detection, pothole detection. A modern vehicle is usually
equipped with a number of sensors, including speedometer,
camera, stability control sensor, suspension system sensor,
etc. The sensor readings from a single vehicle may be
inaccurate or even erroneous. For example, when a vehicle
runs over a pothole, the on-board computer may obtain a
80 percent probability that there is a pothole. Therefore, it is
desired that information from different vehicles can be
merged to increase its accuracy. From the perspective of
applications, it is important to efficiently aggregate raw
data to improve the accuracy of observation.
Function Approximation
In this section, we first discuss the challenge behind our
problem and then propose a fuzzy-rule-based function
approximation method.
Theoretically, we can use any regression method to train
the Q function, but our particular task has several unique
properties. Vehicle densities 1 and 2 have a significant
impact on network connectivity, but it’s difficult to find an
accurate transition threshold; if higher than the threshold,
the network is connected; if lower than the threshold, the
network is disconnected. Similarly, Tw implicitly indicates
how far away the recent aggregatable report is. It’s also
difficult to define what is near and what is far. In general,
it’s difficult to directly define a linear or polynomial model
to represent the relation between the Q function and its
variables 1; 2; Tw; a.
B
CONCLUSION
In this paper, we have presented a data aggregation scheme
for VANETs based on distributed learning. Essentially, the