s of this page or responsible for its content.
Resource Allocation and Outage Control for Solar-Powered WLAN Mesh Networks
Resource Allocation and Outage Control for
Solar-Powered WLAN Mesh Networks
Amin Farbod and Terence D. Todd, Member, IEEE
AbstractIn this paper, resource allocation and outage control are considered for solar-powered WLAN mesh networks. Solar-
powered nodes are a very cost effective option in WLAN mesh deployments where continuous power sources are not practical. In such
nodes, the cost of the solar panel and battery can be a significant fraction of the total and, therefore, reducing AP power consumption is
very important. A solar panel/battery configuration methodology is introduced based on a proposed AP power-aware version of
IEEE 802.11. Public meteorological data is used to provision each node based on an averaged offered capacity profile. Since a node is
configured statistically, it is possible that future loading may result in nonzero outage even when negligible outage is the design target.
Control algorithms are introduced which can improve node outage performance by sometimes introducing an access point capacity
deficit. Results are presented which show the value of the proposed configuration methodology and show that the control algorithms
can prevent outage even at high levels of excess loading.
Index TermsWireless LANs, IEEE 802.11, WLAN mesh, power saving, access point, solar power, multihop communications.
Ç
1
I
NTRODUCTION
W
LAN
mesh networks are used to provide IEEE 802.11
coverage using multihop relaying between mesh APs
(MAPs) and mesh points (MPs). These networks are
currently being standardized under IEEE 802.11s [1], which
intends to promote interoperability between different
vendor solutions. One of the major costs of certain WLAN
mesh deployments is that of providing MAPs/MPs with
electrical power and wired network connections. This is
especially true in Wi-Fi hotzones, where coverage is
provided over extended outdoor areas. Although power
can be supplied through power over Ethernet (POE), such a
solution requires a wired network connection, which is
often very expensive. For the past several years, the
SolarMESH network has been under development and
undergoing deployment trials at McMaster University. In
SolarMESH, the APs are solar-powered and completely
tetherless and can be deployed quickly and inexpensively
for outdoor Wi-Fi coverage in campuses, building com-
plexes, and other Wi-Fi hotzones.
In solar-powered WLAN mesh networks, node resource
allocation includes assigning a panel and battery size to
each mesh node. This assignment is very important since
the panel and battery can be a significant fraction of the
total cost, especially in temperate regions. The assignment
must take into account the power consumption of the node
and, for this reason, it is important to minimize this as much
as possible. Unfortunately, IEEE 802.11 does not include
any native procedures that would allow an access point to
achieve power saving, and this aspect of IEEE 802.11 is an
impediment to the development of a real power saving
WLAN infrastructure. In classical IEEE 802.11, power
saving has dealt with end user stations since access points
are assumed to have continuous power connections [2].
When the solar panel/battery configuration is per-
formed, a load profile for each node is determined. The
load profile is a time function which represents the peak
workload for which the node in question is designed. For a
given geographic location, public meteorological data is
then used to design the node subject to a target outage
probability. In this paper, we assume that the design uses a
proposed power saving mechanism based on simple
extensions to IEEE 802.11. This approach is statistical since
future load conditions and solar insolation may not exactly
match that for which the node was designed. For this
reason, control algorithms are introduced which attempt to
maintain outage-free operation of the node by sometimes
introducing a capacity deficit. Results are presented show-
ing that significant resource reductions are possible using
the proposed design. The control algorithms are also shown
to prevent outages even at high levels of excess loading and
are shown to perform well compared with capacity deficit
lower bounds. The results also give a strong motivation for
including access point power saving in outdoor WLAN
mesh networks.
The remainder of the paper is organized as follows: In
Section 2, we discuss work related to the design and
provisioning of photovoltaic systems relating to networking
applications. In Section 3, we then discuss solar-powered
access point design and modeling based on the energy flow
model. This model, combined with public solar insolation
data, can be used to model the performance of the nodes.
Then, in Section 4, we introduce the solar panel and battery
configuration technique used in the remainder of the paper.
Following this, in Section 5, the outage control problem is
960
IEEE TRANSACTIONS ON MOBILE COMPUTING,
VOL. 6,
NO. 8,
AUGUST 2007
.
A. Farbod is with The Edward S. Rogers Sr. Department of Electrical and
Computer Engineering, University of Toronto, 10 Kings College Road,
Toronto, Ontario, M5S 3G4, Canada. E-mail: afarbod@comm.utoronto.ca.
.
T.D. Todd is with the Department of Electrical and Computer Engineering,
ITB-A324, McMaster University, Hamilton, Ontario, Canada L8S 4K1.
E-mail: todd@mcmaster.ca.
Manuscript received 4 Dec. 2005; revised 28 Apr. 2006; accepted 18 July
2006; published online 15 June 2007.
For information on obtaining reprints of this article, please send e-mail to:
tmc@computer.org, and reference IEEECS Log Number TMC-0360-1205.
Digital Object Identifier no. 10.1109/TMC.2007.1079.
1536-1233/07/$25.00 ß 2007 IEEE
Published by the IEEE CS, CASS, ComSoc, IES, & SPS discussed and two control algorithms are introduced based
on classical control mechanisms. A lower bound on the
achievable outage-free capacity deficit is derived in the
Appendix, and performance results are given in Section 6.
Finally, concluding remarks are made in Section 7.
2
R
ELATED
W
ORK
In a solar-powered WLAN mesh, the MAPs/MPs are
photovoltaic (PV) systems which provide reliable operation
by achieving a sustainable balance between energy input
and output. PV designs have been used in communication
systems for many decades and their applications now range
from powering telephone systems to operating emergency
message centers in rural areas. Several projects have used
solar energy to power IEEE 802.11 relays into otherwise
inaccessible areas [3]. In [4], [5], and [6], solar power is used
in a wireless sensor network. Some nodes will be in a
position to receive more solar energy than others and, thus,
they are better equipped to perform functions such as
packet forwarding. In [7], network lifetime was improved
by extracting environmental energy in another case where
energy supplies are not homogeneous. In [8], solar-powered
wireless detectors are used in an urban traffic information
collection system. In [9], several protocols were proposed
which can achieve mesh AP power saving in cases where
legacy end stations are being accommodated. Unfortu-
nately, there are many practical disadvantages to these
procedures, mainly caused by IEEE 802.11s assumption
that an AP is always active on its assigned channel.
A problem extensively studied in the literature is the
sizing of photovoltaic systems. Different sizing methods
and their accuracies are studied in [10] along with
discussions of how various PV components result in
different system configurations. Three probabilistic meth-
ods for sizing PV systems are compared in [11]. The first
method sizes the battery so that it can support a fixed load
for a set number of days. Similarly, the panel is sized so that
a full battery recharge can be done within a specified time
period. The second scheme is based on detailed computer
simulations. The last approach is based on Markov chain
modeling of the battery state-of-charge. It has been shown
that the second method yields by far the most accurate
results [11] and this is the method we adopt in this paper. In
[12], a Markov chain model was used for the battery state of
charge in PV systems. To use this approach, however, the
variance and mean of the daily solar insolation must be
known. This model was refined in [13] and includes the
effects of interdaily correlation in solar irradiation. Another
performance evaluation of PV systems based on Markov
chain modeling is presented in [14]. Another closed form
solution for PV system sizing is studied in [15].
In [15] and [16], the sky clearance index is used for
simulation of solar irradiance. Borowy and Salameh [17]
consider the optimum PV array sizing for a stand-alone
hybrid wind/PV system. In [18], variable loading is taken
into account in the sizing of PV systems. A fuzzy decision-
making process is used in [19] to evaluate subjective factors
in the PV system sizing decision. Energy management
issues in a space station have also been considered under
limited energy constraints [20].
Much of the quoted work assumes idealized battery
models which can lead to significant inaccuracies. In
practical PV systems, for example, battery capacity is a
strong function of the ambient temperature. In [21], [22],
and [23], some relevant models for the battery behavior
are studied.
3
S
OLAR