Chapter 17: Green Broadband Access Networks
Tao Han, Jingjing Zhang, and Nirwan Ansari
Advanced Networking Laboratory, New Jersey Institute of Technology,
Newark, NJ, United States
HANDBOOK ON GREEN INFORMATION AND COMMUNICATION SYSTEMS
2
Green Broadband Wireless Access Networks
Techniques on greening cellular networks
Power saving communication protocols
Heterogeneous network deployment
Enabling off-grid BSs
Greening via cooperative networking
Cooperation among BSs
Cooperation between BSs and UEs
3
Power saving communication protocols
Idea: Adjusting the transmit power of the transceivers according to the traffic intensity
Traffic volumes variation
Typical day-night behavior of users Mobility of users
Users tend to range over their office districts during working hours and stay home in their residential area after work. This
results in the surge of traffic in both areas at peak usage hours, but in the drop of traffic during the off-peak hours.
SolutionsSwitching off the transceivers when the traffic load is below a certain threshold for a certain time period. When some base transceiver stations are switched off, radio coverage and service provisioning are taken care of by the remaining active devices.
4
Heterogeneous Network Deployment
Disadvantages of Homogeneous Network Deployment Optimization of the location of BSs is complicated Limited ability to adapt to the traffic load.
Heterogeneous Network Deployment Utilizing a diverse set of base stations can be deployed to
improve spectral and energy efficiency per unit area.
5
Enabling Off-Grid BSs (1)
Designing off-grid BSs and communication protocols to enable optimal utilization of renewable energy in cellular access networks
Off-grid BSs
6
Enabling Off-Grid BSs (2)
Designing communication protocols to maximize the utilization of green energy Energy Source Aware Target Cell Selection (Ref. [31])
The proposed algorithm is to ease the mobile users to handover into the green cell and also to make the UE more difficult to leave the green cell. As a result, the coverage of the green cell is actually enlarged, therefore reducing the on-grid power consumption.
7
Greening via Cooperative Networking: Cooperation among BSs
Green opportunities
Cooperation umbrella cell and the underlying cells in multi-layer cellular network architecture
Cooperation among BSs in flat cellular network architecture
Cooperation among BSs from different mobile service providers
8
Greening via Cooperative Networking: Cooperation among BSs
Green challenges
When to cooperate: determine the traffic threshold for cooperation
• If the threshold is too high, the coalition will break down in a short time
period. In other words, some BSs that were turned into the sleep mode
in the cooperation will restart soon. In this case, the energy consumed
by restarting the BSs may be much higher than that of noncooperation.
• If the threshold is too low, the BSs may miss some cooperative
opportunities.
Who to cooperate: determine the coalition among BSs
• Determine the size of the coalition
• Determine the members of the coalition
9
Greening via Cooperative Networking: Cooperation between BSs and UEs
Green opportunities
Unified cellular and ad-hoc network architecture (ref. [33])
Cooperation with beamforming
• Transmit beamforming provides incentives to the relay users to stimulate the
cooperation
• One hop relay to attain largest performance improvement while consuming minimal relay
energy
10
Greening via Cooperative Networking: Cooperation between BSs and UEs
Cooperation network protocols
1. Channel measurement and data
request
2. Transmission strategy calculation
3. Relay selection and cooperation
negotiation
4. Relay assignment and relay
negotiations
5. Relay assignment
acknowledgement
6. Data transmission
11
Greening via Cooperative Networking: Cooperation between BSs and UEs
Green challenges
Channel state information
• New protocols to enable the measurements and updates of the
channel state information among UEs
• How to efficiently feedback the channel state information to BSs
Incentive mechanism
• Design incentive mechanism to avoid tragedy of common
Hybrid handover scheme
• Design an efficient handoff scheme to address the handovers from
BSs to BSs, from BSs to relay UEs, from relay UEs to relay UEs, and
from relay UEs to BSs.
12
Greening via Cooperative Networking: Cooperation between BSs and UEs
Case study: energy efficient wireless multicasting (ref. [34])
The energy efficient wireless multicasting integrates multicast beamforming and cooperative networking. It contains two phases: in phase 1, the base station (BS) transmits the signal to the subscribers using antenna arrays with multicast beamforming; in Phase 2, the users who successfully received the signal in phase 1 forward the signal to other users. The unsatisfied users combine the received signals in both phases to retrieve the infor-mation.
13
Greening via Cooperative Networking: Cooperation between BSs and UEs
Case study: simulation results (1)
The simulation compares the minimal transmit power of different multicasting strategies. With transmit beamforming, BS saves more than 3dBm transmit power. Lozano's algorithm is a multicast beamforming algorithm that does not consider cooperation. As the number of users increases, the performance of the proposed algorithm becomes better because there are more cooperative opportunities. When the number of users is larger than 40, the performance becomes steady, in which it uses about 3.5dBm, 2dBm, and 1dBm less transmit power than those of Lozano's algorithm, respectively. It becomes steady because when the number of users is large enough (40 in the simulation), the cooperation gain is not limited by the cooperative opportunities, and becomes steady.
14
Greening via Cooperative Networking: Cooperation between BSs and UEs
Case study: simulation results (2)
This simulation compare the BS power consumptions under different multicasting strategies. . The blue line indicates the power consumption of the standard LTE Macro BS, which can be considered as the power constraint of BS. Note that simply broadcasting without beamforming and cooperation cannot satisfy the users' requirement under the constraint. As compared to the Lozano's multicast beamforming algorithm, our proposed algorithm can save at least 100 Watts when the number of users is larger than 60. The power savings are benefited from the cooperation between BS and users.
15
Green Broadband Wireline Access Networks
Why saving energy consumption of optical access
network is important?
Where and how much is the power consumed in
Passive Optical Network?
Where is the power wasted?
Optical network unit (ONU) and optical line terminal (OLT)
How to save? - proposals
Vision and challenges
Proposals
16
Motivation
Fiber-To-The-x (FTTX) users are increasing year by year By 2011, there are over 80 million FTTx users
Global broadband subscriber forecast
Power consumption of FTTx networks
Power consumption: On average, each FTTx user consumes ~15w (>30w) In 2011, the total FTTx energy consumption is ~11 TWhr, equal to 7M tons of CO2, 3 extra 500MW power stationsGoals: Save FTTx energy
consumption!
17
Background: Passive Optical Network
residential
residential
Business
Central office
Optical line terminal (OLT) ……
……Distribution
1 OLT chassis • contains 8 OLT line cards;• consumes ~100w 1 OLT line card connects with 32 ONUs OLTs consume <40% of FTTx energy (NTT) 1 ONU consumes ~10w
ONUs consumes >60% FTTx energy
…
…
…
Optical network unit (ONU)
Passive optical network (PON): the major FTTx technology
Question: How to save energy at OLT and ONUs?
18
Upstream and downstream scenarios
Upstream scenario
Pros: the upstream traffic arrival triggers the wakeup of asleep ONUs
Challenges: physical layer implementation, fast wakeup and fast sleep
Downstream scenario
Owing to the broadcast nature, an ONU needs to be awake all the time to check the header of each packet!
ONU userONU
upstream
ONU2
ONU1
ONU3
2 2 3 22
3
2 2 32
23
OLT splitter ONU2
ONU1
ONU3
Control scheme is needed to put an ONU into sleep when it doesn’t have downstream packets!
1. When the upstream buffer is empty for some time, ONU enters into sleep2. Upon the upstream traffic arrival, ONU wakes up
19
How to put an ONU into sleep in the downstream scenario?
Existing proposal: two-way or three-way handshake
When the downstream queue of an ONU is empty for some time, OLT sends a message informing ONU to sleep
ONU sends an ACK to confirm the sleep
It addresses the problem, but
At least one round trip time needs to be taken for negotiation
EPON MPCP protocol needs to be extended to support the mechanism
ONUOLT
Sleep notification
Sleep ack
Ref: J. Mandin,10G-EPON task force meeting 2008, R. Kubo et al., Globecom’09, JOCN’10 S. Wong, et al. Greencom’09, OFC’10
20
Our proposal – main idea Main idea: let ONU infer its downstream queue status
instead of being explicitly notified by OLT
Assume OLT schedule downstream traffic of ONUs with nonempty queues in order (e.g., 1, …, N)
If no traffic is destined to an ONU for some time, the ONU can infer that it doesn’t have downstream traffic, and then go to sleep
Onu 1OLT ONU 1onu2 onu3onu4 onuN… onu2 onu3 onu4 onuN… onu2 onu3 onu4 onuN…
ONU 1 guess it doesn’t have traffic
ONU 1 is more sure it doesn’t have traffic
21
Our proposal
Implement a sleep control algorithm at ONU
How can an ONU avoid missing downstream packet when it is sleeping? Solution: Implement a sleep control algorithm at OLT
Pros: easy implementable, compatible with current protocol
OLT ONU
Sleep control algorithm:If I haven’t sent traffic to ONU i for time t_silent, then, I will buffer its traffic which arrives in the next t_sleep time
OLT ONU
Sleep control algorithm:If no traffic is sent to me for time t_silent, then, I will go to sleep for time t_sleep
22
Energy-Efficient OLT
Peak hourOff-Peak hour
Typical daily traffic profile
Source: Amsterdam Internet Exchange
Off-peak hour traffic rate is much less than peak hour traffic
Current status
One OLT chassis contains multiple line cards All OLT line cards are power-on all the time
OLT traffic profile
Why not aggregate traffic of multiple line cards and power off some line cards in off-peak hour?
23
Energy-efficient OLT OLT
How to aggregate traffic of multiple PONs at OLT?
Assume one OLT chassis contains 4 line cards
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
4×4 optic
al switc
h
OLT chassis Green OLT chassis
ONUs
ONUs
ONUs
ONUs
ONUs
ONUs
ONUs
ONUs
24
Energy-efficient OLT
Assume the switching speed is fast (<1 DBA cycle), i.e., the switch configuration can be rather dynamic
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
Case 1: Load>75% Case 2: Load [50%, 75%]
Case 3: Load<[25%,50%]
OLT line card
OLT line card
OLT line card
OLT line card
Case 4: Load<25%
25
Energy-efficient OLT
Assume the switch speed is slow,
i.e., the switch configuration is semi-static
Further assume the traffic is uniform among all PONs
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
OLT line card
Case 1: Load>50% Case 2: Load [25% 50%] Case 3: Load<25%
26
Simulation results
Conclusions Green broadband wireless networks
Techniques on greening cellular networks• Power saving communication protocols• Heterogeneous network deployment• Enabling off-grid BSs
Greening via cooperative networking• Cooperation among BSs• Cooperation between BSs and Ues• Case study: energy efficient wireless multicasting
Green broadband wireline networks Energy consumption measurement in Passive Optical
Networks• Energy waste in ONU• Energy wast in OLT
Energy efficient Passive Optical Networks• Energy efficient ONU• Energy efficient OLT
28
Thanks for your attention!