Pid967241

Towards Green Broadband Access Networks
Pulak Chowdhury, Massimo Tornatore, Suman Sarkar, and Biswanath Mukherjee
University of California, Davis, CA 95616 USA
Email:{pchowdhury, mtornatore, sumsarkar, bmukherjee}@ucdavis.edu
Abstract—Energy consumption in next-generation access networks networks should be green featuring efficient energy management
is rapidly increasing with the increase of bandwidth demands. In schemes to reduce “carbon footprint.”
this paper, we devise design techniques aimed to build “green” Till now, most of the power management approaches in
access networks to reduce energy consumption. We apply these
techniques on a novel access network paradigm named Wireless- networking research focus on increasing energy efficiency of
Optical Broadband Access Network (WOBAN). We present a Mixed equipments. There are two other directions of network energy
Integer Linear Program (MILP) model to investigate the power management – energy-aware network design and energy-aware
consumption in WOBAN over dynamic traffic profiles. We analyze protocol design [3], [7]. In this paper, we develop energy-aware
the impact of energy-aware design on the performance of WOBAN. design techniques and routing protocol for “green” WOBAN.
Our results indicate that there are large scopes of power savings if
we incorporate energy-aware design and routing in access networks. Although these techniques are developed for WOBAN, we believe
that they will be proven to be more general and can be adapted
I. I NTRODUCTION to other access network technologies like wireless networks and
different PON variants. To the best of our knowledge, this is the
Access network is the “last mile” of the communication net- first paper to devise energy-consumption reduction techniques in
work that connects the telecom Central Office (CO) to the resi- hybrid wireless-optical access networks.
dential and business customers. With the proliferation of the In- The remainder of this paper is organized as follows. In
ternet, customer demands for bandwidth-intensive services (such Section II, we briefly discuss related work on network energy
as Video-on-Demand, online Gaming, HDTV, etc.) are rapidly management. Section III describes the WOBAN architecture,
increasing. Therefore, today’s access network should exhibit techniques for energy-aware WOBAN design, and energy-aware
higher transport capacity. There are several access technologies WOBAN routing protocol. In Section IV, we present a case
proposed and deployed in the market – Digital Subscriber Line study to illustrate the effectiveness of our energy-aware design
(xDSL), Cable Modem (CM), Wireless and Cellular networks, on WOBAN. Section V includes illustrative numerical examples
fiber-to-the-x (FTTx), Hybrid Wireless-Optical Broadband Access on the case study and gives insights on better power management.
Network (WOBAN), etc. Access technologies such as xDSL, CM, Finally, we conclude in Section VI.
Wireless, and Cellular networks do not live up to satisfying future
broadband Internet demands. FTTx technologies can provide II. R ELATED W ORK
higher bandwidth but still remain cost-prohibitive. WOBAN – Many research efforts in network energy management focus
a novel hybrid access network paradigm with the combination on reducing system-level power consumption. In [8], the authors
of high-capacity optical backhaul and wireless front-end – can advocate for the use of optics to increase router’s scaling capacity
provide higher bandwidth in a cost-effective manner [1]. and reduce power consumption. Dynamic power management
Performance of access technologies is improving over time, techniques for system components are presented in [9].
reducing the cost per byte of traffic and making the broadband There has been significant amount of research efforts for power
Internet affordable to more users. This refuels the tremendous management in mobile adhoc networks and sensor networks.
growth of the Internet and scales up the size of broadband access Mobile adhoc networks along with sensor networks are con-
networks. For example, let us consider FTTx which has different strained with limited power capabilities. Hence, these networks
underlying technologies, such as direct fiber, shared fiber, and and their protocol design should emphasize on power manage-
the most dominant one – Passive Optical Network (PON) (e.g., ment techniques. Reference [10] provides a nice overview of
EPON, GPON, etc.). Infonetics Research [2] – a leading market these techniques at both MAC and network layers. At MAC layer,
research firm – estimates that the worldwide PON equipment some proposed protocols focus on algorithms to put the radios of
market grew 17% during the second quarter of 2008. the idle wireless nodes in coordinated sleeping. Other proposals
With the increase of size and bandwidth demand in the In- include TDM link scheduling, protocols to minimize unnecessary
ternet, power consumption also increases. An estimation shows transmission, etc. [4]. The network layer protocols focus mainly
that the Internet electricity consumption at US is in billions of on selecting routes to minimize network energy consumption.
dollars [3], and there is significant wastage of electricity (several Data centers and server clusters consume a great deal of
TWh/year) due to inefficient network and system design [4]. A electricity power for searching, cooling, and other purposes. A
good portion of this energy is consumed by idle network elements large number of scholarly papers address the power management
[3]. Therefore, energy can be conserved by reducing the power issues of data centers and individual servers (for example, [11]).
consumption of these idle network elements. Researchers are also arguing to relocate data centers near the
Access network comprises a large part of the Internet. It is renewable energy sources to build “green” grid technology. In
also a major energy consumer in the Internet due to the presence [6], the authors make a proposal for designing “follow-the-sun,
of huge number of active elements [5]. If we can reduce energy follow-the-wind” grids to build green data centers.
consumption in the access network, it will automatically provide Authors of [7] advocate power-aware design and configuration
a significant reduction of overall Internet power consumption. of core networks. They use optimization techniques to find
Access network power-consumption reduction not only has the minimal configuration of the router chassis for supporting differ-
potential of enormous cost savings, this will also lead us to ent traffic loads, thereby reducing network power consumption.
develop environment-friendly technologies, thereby achieving the An initial proposal on the required architectural change of the
ultimate goal of “green” Internet [6]. Therefore, future access network to support selective connectivity, a state where a host
978-1-4244-4148-8/09/$25.00 ©2009
This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE "GLOBECOM" 2009 proceedings.

can choose whether to stay connected or disconnected, is given in providing high level of performance. Thanks to the mesh front-
[12]. Selective connectivity state happens when hosts go to sleep end, traffic can be rerouted through alternate paths in case of
to reduce network power consumption. Methods which enable failures such as fiber cut, wireless router or gateway or ONU
energy consumption reduction of networked desktop computers failures. Moreover, there is a capacity mismatch between the
while these computers are idle are presented in [3]. Authors of [4] wireless front-end and optical backhaul. The redundant capacity
suggest how coordinated and uncoordinated sleeping mechanisms in the optical backhaul will provide extra reliability during the
can put the Internet nodes to sleep (energy saving state) with failure so that traffic can be rerouted through alternate paths. At a
some changes in Internet protocols. Despite all these efforts, there specific instant, it is possible to find several WOBAN topologies
remain significant challenges to deploy energy-efficient schemes that can satisfy the required capacity and reliability objectives.
over the Internet, especially over the access networks. All these are possible due to the densely interconnected wireless
mesh front-end which has many redundant paths to route traffic.
III. G REEN WOBAN
The flexibility provided by the wireless front-end of WOBAN can
In this section, we present the architecture of WOBAN, a be exploited to enable energy savings in the optical part.
mathematical model of energy-aware WOBAN design, and an There is another important observation on access network
energy-aware routing algorithm for WOBAN. traffic profile. The traffic load on access network comes directly
A. WOBAN Architecture from customers, and it is well known that there are daily
fluctuations of this load. During WOBAN (as well as other
Hybrid Wireless-Optical Broadband Access Network access networks) deployment, the common practice is to deploy
(WOBAN) is a novel access network architecture with an network equipments so that it can support maximum traffic load.
optimal combination of optical backhaul (e.g., a Passive Optical Consequently, during low-load hours, some parts of the network
Network (PON)) and a wireless front-end (e.g., WiFi and/or may be under-utilized.
WiMAX). Figure 1 presents the architecture of WOBAN. Hence, to design WOBAN topologies with reduced power
WOBAN optimizes the deployment cost due to less-expensive consumption, we need to consider the following points – (a)
wireless front-end and maximizes the bandwidth performance of deployed network has some extra capacity, and (b) traffic load
a broadband access network [1]. varies during different hours of the day. Thus, we can selectively
put some nodes to a low-power (sleep) state during low-load
hours, thereby reducing network-wide power consumption.
In the wireless front-end of WOBAN, we can adopt coordinated
sleeping techniques [4] from mobile adhoc networks research to
reduce wireless router energy consumption. Therefore, in this
paper, we mainly focus on how to put optical components of
WOBAN into sleep state. We will not consider putting OLT into
sleep state as it connects the WOBAN to the rest-of-the-Internet.
However, for protection purposes, in a PON segment, it is possible
to have several OLTs in a ring setup. In that case, a low-load OLT
can be put into sleep state while rerouting its traffic through other
OLTs. In this paper, we intend to reduce ONU power consumption
in WOBAN by putting low-load ONUs into sleep state.
Now, how can we put an ONU to sleep state? Current IEEE
Fig. 1. WOBAN architecture 802.3ah/ 802.3av standards do not define any low power state for
In WOBAN (Fig. 1), a PON segment starts from the Optical ONU [13]. However, proposals have been made to IEEE 802.3av
Line Terminal (OLT) at the telecom CO and ends at multiple task force to include low-power states for ONU so that it can
Optical Network Units (ONU). Multiple wireless routers form the go to sleep during idle periods [14]. Typical power consumption
front-end of WOBAN. A selected set of these routers are called by an ONU during active state is approximately 10 W [15]. It
gateways. The front-end of WOBAN is essentially a multi-hop is also estimated that during sleep state, power consumed by an
Wireless Mesh Network (WMN) with several wireless routers ONU is less than 1 W [14]. Existing ONU boxes in the market
and a few gateways. These gateways are connected to the PON include a T X DISABLE input which disables the transceiver
backhaul through the ONUs. Each ONU can support several of an ONU [15]. Disabling the transceiver can reduce ONU power
wireless gateways. End users (both mobile and stationary) connect consumption several fold.
to WOBAN through the wireless routers. In WOBAN, OLT can manage a centralized sleeping mecha-
nism to put low-load ONUs into sleep. The mechanism works as
B. Energy-Aware WOBAN Design follows. An OLT maintains two watermarks for the traffic load
WOBAN represents a hierarchical access architecture with at ONUs – low and high watermark. At OLT, there are input
gateways as the initial traffic aggregation points. ONUs are the queues for each ONUs connected to it. During different hours of
next aggregator level in the hierarchy, while OLT is the highest the day, OLT will observe the traffic load at different ONUs by
aggregation point and connects the access network with the measuring the length of corresponding input queues. If an ONU
metro/core network. As gateways are part of the WOBAN front- is operating under low watermark, OLT can put that ONU into
end, they can also collect their nearest end-user traffic. sleep. The wireless mesh front-end of WOBAN will reroute the
There are several important aspects of WOBAN we need affected traffic due to ONU shut-down to alternate paths. OLT
to consider for energy-aware design. Current WOBAN design, can put back sleeping ONUs into active state when traffic load
deployment, and management idealogies provide infrastructure increases above high watermark in the currently active ONUs.
that showcases fault tolerance, reliability, and robustness while Now, we aim to find out the optimal number of ONUs needed
978-1-4244-4148-8/09/$25.00 ©2009

to support a given amount of traffic load. This is a form of multi- +Zs,d , u = s
(u,v) Zs,d γu,v −
s,d
(v,u) Zs,d γv,u =
s,d
−Zs,d , u = OLT
commodity flow problem where each commodity represents the (3)
0, otherwise
traffic flow between a source-destination pair. We can formulate ∀u ∈ V, ∀(s, d) ∈ T
the problem as a Mixed Integer Linear Program (MILP). In the Wireless Capacity Constraint: A wireless link in WOBAN can
following, we present the mathematical model in MILP form. carry both upstream and downstream traffic. Equation 4 states that
1) Mathematical Model: Our model takes as input a WOBAN the summation of all traffic through a wireless link (u, v) should
not exceed the capacity (Cu,v ) of the link.
with preassigned link capacities and a traffic matrix based on
the dynamic daily traffic profile of a service area. The model (Vs,d λs,d + Zs,d γu,v ) ≤ Cu,v ,
u,v
s,d
∀(u, v) ∈ Ew (4)
generates output which determines minimum number of ONUs (s,d)
that need to be kept active to route the given traffic load. The Wireless Constraints: The dual-threshold interference model
other ONUs can be put to sleep state so that network-wide power [16] is used to find the set of all interfering links at each
consumption is minimized. The model also finds the routing path wireless node of WOBAN. The wireless interference constraints
are translated to constraints that allocate capacity on each wireless
for each source-destination (s, d) pair in the traffic matrix. In link. A wireless node divides its capacity to all its incoming and
WOBAN, for the upstream traffic, the destination is always OLT outgoing links as it can not transmit and receive at the same
where the source can be one of the wireless routers, and for the time. So, the interference-free radio capacity available (Cu ) at
downstream traffic, the source is the OLT and the destination is each node u is shared between all the outgoing links from u
one of the wireless routers. (first term in Eqn. 5) and all the incoming links to u (second
term in Eqn. 5).
To describe the model, we introduce some notations for the
parameters and variables as follows. Cu,v + Cv,u ≤ Cu , ∀u ∈ Vw (5)
v v
• WOBAN topology: denoted by a weighted and directed
Equation 6 forms the secondary interference constraint of the
graph G = {V, E} where V is the set of nodes and E is the wireless mesh of WOBAN. This constraint states that a node
set of links. V has three subsets – Vw is the set of wireless cannot receive any signal from any other node when an interfering
nodes, Vonu is the set of ONUs, and OLT represents the link is active. The first term of Eqn. 6 is same as Eqn. 5,
OLT. If nodes u and v have a link, the link is denoted by representing the shared capacity among incoming links to node
u. The second term represents all the links which interfere with
(u, v). E has several subsets – Ew is the set of wireless node u (Iu,v ) and which do not have node u as one of their end
links, EOG is the set of ONU-to-Gateway links, EGO is the points.
set of Gateway-to-ONU links, ET O is the set of OLT-to-
Cv,u + Cp,q ≤ Cu , ∀u ∈ Vw (6)
ONU links, and EOT is the set of ONU-to-OLT links.
v (p,q)∈Iu,v
• (s, d): identifies source-destination pair of the
Wired Capacity Constraints: Downstream traffic flows are
upstream/downstream traffic in the traffic matrix. limited by the capacity of the ONU-to-GW (Eqn. 7), and OLT-to-
• Xu : Binary variable denoting ONU state, Xu ∈ {0, 1}. 0 ONU (Eqn. 8) links. Similarly, upstream traffic flows are limited
denotes ONU is asleep, and 1 denotes ON U is active. by the capacity of the GW-to-ONU (Eqn. 9), and ONU-to-OLT
s,d
• λu,v : Binary variable denoting download flow on link (u, v) (Eqn. 10) links.
for a (s, d) (s is OLT, d denotes routers) pair, λs,d ∈ {0, 1}.
u,v Vs,d λs,d ≤ COG ,
u,v ∀(u, v) ∈ EOG (7)
s,d
• γu,v : Binary variable denoting upload flow on link (u, v) for (s,d)
a (s, d) (s denotes routers and d is OLT) pair, γu,v ∈ {0, 1}.
s,d
Vs,d λs,d ≤ CT O ,
u,v ∀(u, v) ∈ ET O (8)
• Cu,v : Variable expressing the capacity over a wireless link (s,d)
(u, v). This variable can assume non-integral values. s,d
Zs,d γu,v ≤ CGO , ∀(u, v) ∈ EGO (9)
• COG : Capacity of ONU-to-Gateway link. (s,d)
• CGO : Capacity of Gateway-to-ONU link. s,d
Zs,d γu,v ≤ COT , ∀(u, v) ∈ EOT (10)
• COT : Capacity of ONU-to-OLT link. (s,d)
• CT O : Capacity of OLT-to-ONU link. Wired Directionality Constraints: These constraints ensure that
• T : Input traffic matrix with two different types of traffic no upstream traffic is flowing in the downstream direction in the
values - (1) Vs,d : Download traffic between a (s, d) pair wired part of WOBAN and vice versa.
and (2) Zs,d : Upload traffic between a (s, d) pair. For each λs,d = 0,
u,v ∀(u, v) ∈ EOT ∪ EGO , ∀(s, d) ∈ T (11)
(s, d) pair, we assume the upload traffic is a fraction of s,d
Vd,s γu,v = 0, ∀(u, v) ∈ ET O ∪ EOG , ∀(s, d) ∈ T (12)
the download traffic, i.e., Zs,d = ft , ∀s,d where ft is
a constant value. ONU State Constraint: This constraint determines the ONU
state. If some traffic (upstream (γu,v ) or downstream (λs,d )) flows
s,d
u,v
Now, the objective function can be written as: through an ONU u, it should be active (Xu = 1), otherwise it
should be in sleep state (Xu = 0). This condition can be expressed
minimize Xu (1) by the following constraint,
u∈Vonu

subject to the following constraints: v (s,d) λs,d +
u,v v (s,d)
s,d
γu,v
Xu ≥ , ∀u ∈ Vonu (13)
Flow Constraints: Equation 2 captures the fact that, in all nodes M
of WOBAN, total outgoing downstream traffic should be equal to where M is a very large value used to map the flow variables
total incoming downstream traffic except for the source (OLT) and
the destination nodes (wireless routers). Similar argument holds into a binary variable (Xu ).
for upstream traffic (Eqn. 3) except for the source nodes (wireless Path Length Constraints: These constraints put a limit on path
routers) and the destination (OLT). length. As Eqns. 14 and 15 shows, each upstream or downstream
(s, d) path should not be longer than H hops.
−Vs,d , u = d
(u,v) Vs,d λs,d −
u,v (v,u) Vs,d λs,d =
v,u +Vs,d , u = OLT λs,d ≤ H,
u,v ∀(s, d) ∈ T (14)
(2)
0, otherwise u,v
∀u ∈ V, ∀(s, d) ∈ T

978-1-4244-4148-8/09/$25.00 ©2009

s,d
γu,v ≤ H, ∀(s, d) ∈ T (15) B
3
C
u,v 3 3
This model turns out to be a MILP formulation as some 7 6
S H D
variables (such as Cu,v ) can take non-integral values. Generally,
MILP problems are known to be NP-hard, but sophisticated 2
1
G
heuristics can be used to solve the problems in a reasonable E F
2 2
amount of time. The number of design variables in this problem
is not exceptionally large. But for a very large number of nodes,
Fig. 2. Residual capacity as link weights.
the problem becomes quite large.
We try to solve our model with smaller networks to verify
To achieve this, we can modify the LS routing algorithm in
the correctness of our formulation. For solving the MILP, we
WOBAN so that link weights are assigned to satisfy our energy
use ILOG CPLEX [17] on a Intel Core 2 Duo machine with 1
objective. So, we use residual capacity as the link weight. Initially,
Gigabyte RAM and Ubuntu Linux OS. CPLEX uses state-of-the-
each link will have its initial capacity as the weight. When traffic
art branch-and-cut for solving MILP problems. We use a small
flows through a link, its next link weight will be the capacity left
43 node WOBAN with 30 wireless routers, 8 gateways, 4 ONUs,
(original capacity minus traffic flow) on that link. To route traffic
and 1 OLT. Each ONU can drive 2 gateways. In this network, the
from source to destination, we find the lowest residual capacity
MILP can be solved in a reasonable amount of time depending
path. For example, let us consider the small network in Fig. 2.
on the size of the traffic matrix. With various combinations of
The links are annotated with their weights (residual capacities).
the load size and path length constraints, the results show that
To send traffic from S to D, energy-aware routing algorithm will
we have to keep at least one ONU alive and in the extreme case,
route traffic through the path S-E-F-G-D which has the lowest
all ONUs are on.
residual capacity (2 + 2 + 2 + 1 = 7).
However, for WOBAN with a larger number of nodes and
This approach, however, has its own shortcomings as shown
more traffic load, we need to build some heuristics to solve this
in Fig. 2. The algorithm selects the path with 4 hops although
problem. We already have a heuristic on deciding which ONUs
that is not the shortest path while using other metrics (such as
to shut down “put ONUs with load less than low watermark to
hop length or delay). This will increase the average path length
sleep.” For routing, we need to find an energy-aware heuristic
and path delay in the network. To deal with this problem, we can
which is described below.
introduce a term named hop offset – the purpose of this term is
C. Energy-Aware Routing
to reduce average path length. If we have a hop offset m, we add
There are several routing protocols proposed for WOBAN- m to the path cost for each hop, i.e., for a path of n hops, the
like architectures. Two of them are – (a) Delay Aware Routing cost of the path will be residual capacity of the path +n × m.
Algorithm (DARA) [18], (b) Capacity and Delay Aware Routing For example, as in Fig. 2, for a hop offset 1, the path costs are
(CaDAR) [18]. These routing algorithms are Link State (LS) 7 + 4 × 1 = 11, 9 + 3 × 1 = 12, and 13 + 2 × 1 = 15 for paths S-
protocols where a node periodically transmits its link-state infor- E-F-G-D, S-B-C-D, and S-H-D respectively, and S-E-F-G-D will
mation to the network by Link-State Advertisement (LSA). Upon be the selected path. But, for a hop offset 3, S-B-C-D will be the
receiving the LSAs from all the nodes, each node finds a map chosen path in our algorithm, and for a hop offset 5, S-H-D will
of the network and can build a routing table (generally by using be the chosen path. Selecting the optimal hop offset depends on
some variant of Dijkstra’s algorithm) to route traffic to other nodes how much delay the network connections can tolerate.
in the network. LS protocols generally vary on how they assign There is another important item to consider. We should select
link weights in the LSA. For example, DARA uses predicted hop offset in such a way that average path length does not
link delay metric to assign link weights. Based on link weight increase unproportionately from regular shortest path routing. If
assignment, these protocols try to achieve several performance average path length increases too much, that means more wireless
objectives. One such objective is load balancing which balances hops per path, i.e., more wireless transmissions and receptions.
the traffic load in all parts of the network [18]. Each wireless transmission/reception requires power. So, out of
Load balancing is a good performance objective as it tries proportion average path length may diminish the power savings
to fairly utilize all parts of the network. But, it may lead to that we gain from putting the ONUs to sleep.
under-utilization of some segments of the network during low-
load hours. During low-load hours, traffic can be supported IV. C ASE S TUDY
using small number of devices in the network. Our routing Figure 3 shows a hypothetical WOBAN deployment scenario
algorithm is an energy-aware LS protocol with the objective in Davis, which is a small city in Northern California near
to reduce network-wide energy consumption by putting under- Sacramento. Davis is the home of the University of California,
utilized nodes (mainly ONUs) of the network into sleep. When Davis. The selected part of Davis has three different areas: (a)
routing traffic, the objective will be “use the already-used paths.” Downtown, (b) UC Davis Campus, and (c) Part of residential
The routing algorithm tries to route traffic through the ONUs area. These areas are selected as they have a very nice blend of
which are already used so that probabilities of other ONUs being technology-savvy users, and we can showcase how traffic profile
unused increase. In that way, we can put zero-load ONUs to varies across different parts of the network, and also depending of
sleep. Moreover, we may find some other ONUs with very low the traffic profile during different time of the day, how we can put
load (ONUs with loads under low watermark). By being more nodes to sleep. The telecom CO is situated in the downtown area.
aggressive, we can put these ONUs to sleep and let the wireless Total 140 wireless routers, 24 gateways, 12 ONUs, and 1 OLT
mesh reroute the traffic through these ONUs. When traffic load are deployed in this hypothetical deployment. The OLT drives 12
increases, sleeping ONUs can be activated to carry increased ONUs, and 1 ONU drives 2 gateways. Downtown has 41 routers,
traffic. [Note: Due to space constraints, we only present the logic 8 gateways, and 4 ONUs; Campus area has 37 routers, 6 gateways,
behind the routing algorithm] and 3 ONUs; and Residence area has 62 routers, 10 gateways,
978-1-4244-4148-8/09/$25.00 ©2009

(c) Part of Davis Residential Area Map Courtesy: Google Maps
1
Downtown Area
Legends: Campus Area
0.9 Residence Area

Telecom CO 0.8

Percentage of Active Routers
OLT 0.7

Splitter
0.6

ONU
0.5
Wireless Gateway
0.4
Wireless Router
Optical Fiber 0.3

CAT-5 Cable 0.2

0.1

0
1
01−03 2
04−06 3
07−09 4
10−12 5
13−15 6
16−18 7
19−21 8
22−24
Hours

(a) Percentage of active routers

3
Downtown Area
Campus Area
Residence Area
2.5

Average Load at Active Routers (Mbps)
(b) UC Davis Campus
(a) Davis Downtown
2

1.5
Fig. 3. Hypothetical WOBAN deployment in Davis.

and 5 ONUs deployed. The routers are equipped with one radio
1

with a capacity of 54 Mbps (IEEE 802.11g). Capacity allocation
0.5
among wireless links is accomplished by TDM link scheduling.
The OLT and ONU have capacities of 1 Gbps and 100 Mbps
respectively. The low watermark is set to 5%, i.e., the OLT puts 0
1
01−03 2
04−06 3
07−09 4
10−12 5
13−15 6
16−18 7
19−21 8
22−24

ONUs with load less than 5% of the total traffic to sleep. Hours

The aforementioned MILP model is not able to produce
optimization results for a large WOBAN like the hypothetical (b) Average load on active routers
deployment in Davis in a reasonable amount of time. Therefore,
we use two heuristic approaches in our simulation to achieve Fig. 4. Traffic profile of Davis Internet users.
near-optimal solution. They are - (a) put ONUs with load less 60
Regular Mode
Power Save Mode
then low watermark to sleep, (b) energy-aware routing algorithm. Energy−Efficient Routing
50
A. Traffic Modelling
To show numerical examples on the performance of energy-
% of ONUs in Sleep State

40
aware WOBAN design, we need to develop a reasonable traffic
profile of the deployment areas during different hours of the day.
30
Access networks deal directly with the user-generated Internet
traffic. So, the behavior of end users has a significant impact on
the performance of the access network. Based on different sources 20

[19] and our observations, we develop a traffic profile for each
of the three different areas of Davis. Each of these areas has 10
different types of Internet users with different behavior patterns
and different peak usage periods. 0
We divide the 24 hours of a day into 8 periods: (1) Period 1 - 1
01−03 2
04−06 3
07−09 4
10−12 5
13−15 6
16−18 7
19−21 8
22−24
Hours
(01-03 hours), (2) Period 2 - (04-06 hours), (3) Period 3 - (07-09
hours), (4) Period 4 - (10-12 hours), (5) Period 5 - (13-15 hours), Fig. 5. Power savings in energy-aware WOBAN.
(6) Period 6 - (16-18 hours), (7) Period 7 - (19-21 hours), and
(8) Period 8 - (22-24 hours). Period 1 begins at 00.01 AM in on these active routers. Both of these data together generate the
the morning. The granularity of these periods can be modified traffic profile of these areas. We assume for each active user, the
as necessary. Figure 4 shows such a traffic profile for downtown, upload traffic is approximately 1 of the download traffic.
4
campus, and residential areas of Davis. There are two parts of
the traffic profile – Fig. 4(a) presents the percentage of active V. I LLUSTRATIVE N UMERICAL E XAMPLES
routers (which is directly proportional to the active users) in these We apply our energy-aware design and routing protocol on the
areas, and Fig. 4(b) shows the average (poisson distributed) load hypothetical WOBAN deployment in Davis (Fig. 3). We collect
978-1-4244-4148-8/09/$25.00 ©2009

5
The average path lengths in the energy-aware routing mode are
Regular Mode
Power Save Mode comparable with the regular mode. Hop Offset tries to reduce the
4.5 Energy−Efficient Routing
average path length in energy-aware routing mode. Otherwise,
4 higher average path length could diminish our power conserving
3.5
benefits with the extra wireless transmission/reception power
consumed. We can see that putting ONUs to sleep does not
Average Path Length

3
significantly increase the average number of wireless hops, thanks
2.5 to the availability of various similar-cost paths in the wireless
2
mesh of WOBAN. Hence, the energy usage in the wireless part
does not increase significantly when we put ONUs to sleep.
1.5
Figure 6(b) provides the average transmission delay of the paths
1 for different setups. Again, the transmission delays are very much
0.5
comparable in all these setups. Consequently, we can state from
the results that we can save a good portion of energy consumption
0
1
01−03 2
04−06 3
07−09 4
10−12 5
13−15 6
16−18 7
19−21 8
22−24 in WOBAN by careful design and energy-aware routing without
Hours compromising the performance.
VI. C ONCLUSION
(a) Average path length
In this paper, we showed that energy consumption of WOBAN
can be reduced by efficient design and energy-aware routing. We
1
Regular Mode
examined the impact of these energy-aware design decisions on
0.9
Power Save Mode
Energy−Efficient Routing
the performance of the network. It appears that with suitable
design decision parameters, we can achieve comparable perfor-
0.8
mance (of WOBAN) with the energy-aware mode. The energy-
saving in the optical part of WOBAN also does not increase the
Average Path Delay (msec)

0.7

0.6 energy usage in the wireless part. We conclude that the energy-
aware design techniques applied on WOBAN can be generalized
0.5
so that they are also applicable to other access networks.
0.4 R EFERENCES
0.3 [1] S. Sarkar, S. Dixit, and B. Mukherjee, “Hybrid Wireless-Optical Broadband
Access Network (WOBAN): A review of relevant challenges,” Journal of
0.2
Lightwave Technology (JLT), vol. 25, no. 11, pp. 3329–3340, Nov. 2007.
0.1
[2] “Infonetics research.” http://www.infonetics.com/, 2009.
[3] K. J. Christensen, C. Gunaratne, B. Nordman, and A. D. George, “The
0 next frontier for communications networks: power management,” Elsevier
1
01−03 2
04−06 3
07−09 4
10−12 5
13−15 6
16−18 7
19−21 8
22−24
Computer Communications, vol. 27, pp. 1758–1770, Aug. 2004.
Hours [4] M. Gupta and S. Singh, “Greening of the Internet,” ACM SIGCOMM’03,
Karlsruhe, Germany, p. 1926, 25-29 Aug. 2003.
[5] C. Lange and A. Gladisch, “On the energy consumption of FTTH access
(b) Average path delay
networks,” OFC/NFOEC’09, San Diego, Mar. 2009.
[6] B. St. Arnaud, “CANARIE.” http://www.canarie.ca/, 2009.
Fig. 6. Energy-aware WOBAN performance. [7] J. Chabarek, J. Sommers, P. Barford, C. Estan, D. Tsiang, and S. Wright,
“Power awareness in network design and routing,” IEEE INFOCOM’08,
Phoenix, Arizona, pp. 1130–1138, 2008.
results for three different setups: (a) WOBAN in regular mode [8] I. Keslassy, S. Chuang, K. Yu, D. Miller, M. Horowitz, and O. Solgaard et al.,
with no energy-saving techniques, (b) WOBAN in power-save “Scaling Internet routers using optics,” ACM SIGCOMM’03, Karlsruhe,
mode (with regular shortest path routing) where low-load ONUs Germany, Aug. 2003.
[9] L. Benini, A. Bogliolo, and G. D. Micheli, “A survey of design techniques
are put to sleep, and (c) WOBAN in energy-aware routing mode for system level dynamic power management,” IEEE Transactions on VLSI
where energy-aware routing is deployed on top of power-save Systems, vol. 8, no. 3, pp. 299–316, Jun. 2000.
mode configurations. We compare the energy savings and measure [10] C. Jones, K. Sivalingam, P. Agrawal, and J. Chen, “A survey of energy
efficient network protocols for wireless networks,” Wireless Networks, vol. 7,
the impact of ONU shutdown on the performance of the network. no. 4, no. 4, pp. 343–358, 2001.
Figure 5 shows the power savings (in terms of percentage of [11] J. Chase and R. Doyle, “Balance of power: energy management for server
clusters,” 8th Workshop on Hot Topics in Operating Systems, May 2001.
ONUs in sleep state) during different periods of the day. Obvi- [12] M. Allman, K. Christensen, B. Nordman, and V. Paxson, “Enabling an
ously, there is no energy saving in the regular mode. Interestingly, energy-efficient future internet through selectively connected end systems,”
on an average, we can put 50% of the ONUs to sleep state in our ACM SIGCOMM HotNets ’07, Atlanta, Georgia, Nov. 2007.
[13] G. Kramer. Personal Communication, Mar. 2009.
hypothetical deployment by using the other two setups. One may [14] J. Mandin, “EPON powersaving via sleep mode.” IEEE P802.3av
argue that if we can put 50% of the ONUs to sleep, why do we 10GEPON Task Force Meeting, www.ieee802.org/3/av/public/2008-09/3av-
deploy them in the first place? The answer is that, at high load, 0809-mandin-4.pdf, Sept. 2008.
[15] “NEC GE-PON data sheets.” http://www.nec.co.jp/, 2009.
when all the routers are active, we will not be able to put any [16] V. Ramamurthi, A. S. Reaz, and B. Mukherjee, “Optimal capacity allocation
ONU to sleep. The power-saving opportunity lies somewhere else in wireless mesh networks,” IEEE Globecom’08, New Orleans, Dec. 2008.
– specifically in the traffic profile. When one part of the network [17] “ILOG CPLEX: High-performance software for mathematical programming
and optimization.” http://www.ilog.com/products/cplex/, 2009.
is at high load, the other parts may be in low load. We can save [18] S. Sarkar, P. Chowdhury, S. Dixit, and B. Mukherjee, Broadband Ac-
energy by putting ONUs in those low-load parts to sleep. cess Networks: Technologies and Deployment, ch. Hybrid Wireless-Optical
Broadband Access Network (WOBAN). Springer, 2009.
Figure 6(a) presents the average path length in three different [19] J. Farber, S. Bodamer, and J. Charzinski, “Measurement and modelling of
setups of WOBAN. In WOBAN, all (s, d) paths have two wired Internet traffic at access networks,” EUNICE’98, Munich, pp. 196–203, Aug.
(OLT⇔ONU, ONU⇔GW) hops, and the rest are wireless hops. 1998.

978-1-4244-4148-8/09/$25.00 ©2009

Pid967241

More Related Content

What's hot (20)

Viewers also liked (10)

Similar to Pid967241 (20)

More from Shahab Shahid (15)

Pid967241