Green Telecom & IT: Vinod sharma : Green Telecom

Preview:

DESCRIPTION

 

Citation preview

Fundamental Limits for Communication Systems withRenewable Energy Sources

Vinod SharmaDept of Electrical Communication Engineering,

Indian Institute of ScienceBangalore, India

Joint work with Utpal Mukherji, R. Rajesh, Vinay Joseph, P. Viswanathand Deekshith K

April 5, 2012

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 1 / 27

Outline

Introduction

Green Communications

Green Communications in India

Communication system design with renewal energy

Single node: Point to PointInformation theoreticQueuing theoretic

MAC

Multihop

Conclusions

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 2 / 27

Introduction

2% of total electrical energy globally consumed in data centers andcommunication equipment.

Predominant ICT energy consumed by Wireless networks.

BS consumes 50% of overall power consumed in wireless networks.

One BS consumes 2Kwatt

50− 80% to RF5− 15% SP10− 25% Air conditioner

A medium sized (12− 15K cell sites) cellular network consumesequivalent of 1, 70, 000 homes.

Every year 1, 20, 000 new BSs added world wide.

Total energy consumed by one cell phone is 0.1Watt.

Manufacturing and disposal of cell phones consume similar amount.

This will have significant environmental impact.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 3 / 27

Green Communication

Saving Energy for specific throughput and QoS satisfaction.

Energy saving should be done at each level

Chip level (hardware): different operating power saving modes, carefulcircuit design.

Energy efficient RF: More efficient design of power amplifiers, savingpower leakage in transmission to antenna.

Software: Operating system, compiler design.

Phy layer: power control, AMC.

Power saving modes under low utilization.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 4 / 27

Green Communication

MAC: Power aware scheduling.

Cross layer design: Phy, MAC, routing.

Smarter design of topology: Cell sites, BS size saves upto 40% energy

Femto cells: Decreases BS and cell phone transmit power

MIMO antennas: to increase capacity, diversity.

Interference coordination

Spectral reuseOpportunistic scheduling

Energy efficient router design.

Should have smaller buffers: Buffer consumes 12 of board space and 1

3of power.

Energy efficient TCP

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 5 / 27

Green Communication

Alternative Energy sources:

Solar

Wind

Fuel cell, hybrid

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 6 / 27

Solar/Wind powered Base Station

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 7 / 27

Indian Cellular Scenario

2, 50, 000 Telecom towers

70% energy consumed by Towers

50% Towers in rural India

Each tower consumes 1K- 3KWattOperated by Diesel generators

≥ 4400 ton/hr CO2 emission.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 8 / 27

Green Communications in India

Solar powered village BS forGSM- World GSM by VNL

Low cost, low powered BS

Connected to large BS

VBS handles hundreds of users

Uses ∼ 100 Watt

2− 8m2 solar panels required

50 VBS installed in Rajasthan

Can provide connectivity atremote places with no electricsupply: saves on diesel.

Other makes: Alcatel- Lucent,Ericsson, Nokia Siemens.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 9 / 27

Green Communications in India

Not enough sun light in Monsoon.

Combining Sun and Wind Energy a solution.

Flexenclosure design of BS

Wind generator atop the tower supporting antenna.

Solar panel on roof of shelter housing switching equipment.

Initial installation cost more than traditional BS but operating costmuch less

No oil/ diesel.No transport cost of oil.Low maintenance.

New TRAI Recommendation: 50% of rural and 20% of Urban BS touse hybrid power by 2015.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 10 / 27

System design with Energy Harvesting sources

Using Solar, wind energy tosupplement regular electricsupply can be effective

BS with Energy Harvestingsources

Downlink

Given (Xk , hk ,Ek) find PK andthe queue to serve so as tosatisfy QoS of different users.

Pk ≤ EK (1) + Ek(2)

Uplink problem

cell phones having solar cells.

Key Concern: Unpredictable, random energy generation.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 11 / 27

Our Research on Energy Harvesting CommunicationSystems

Single Node

AWGN channel with var σ2.

{Yk}iidRK = received from channel =

√TkXk + Wk

Wk ∼ N(0, σ2)

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 12 / 27

Single Node: Capacity

Theorem:

When energy is consumed only in transmission,

the capacity = 0.5log(1 + E [Y ]/σ2)

Comments

1 Limiting capacity achieving distribution is iid N(0,E [Y ])

2 Capacity is same as that of an AWGN channel with average powerconstraint E [Y ].

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 13 / 27

Capacity with Processing Energy

Zk = energy spent in processing and computations.∗{Zk}iid .

Capacity achieving dist. Gaussian iid with possibly sleep mode.

Figure: Capacity with sleep mode.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 14 / 27

Single Node with Data Buffer

{Xk} stationary, ergodic

{YK} stationary, ergodic

TK = min(EK ,E [Y ]− ε), ε > 0 (1)

Theorem:

If E [X ] < g(E [Y ]− ε), g cont., non decreasing, concave then data queueis stable.

(1) is throughput optimal policy.But it is not delay optimal

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 15 / 27

Single Node

Greedy Policy

Tk = min(Ek , g−1(qk)) (2)

Theorem

If E [X ] < E [g(Y )] and energy buffer is finite, then under (2) data queueis stable.

Theorem

If g is linear then (2) is delay optimal and throughput optimal.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 16 / 27

Figure: Comparison of policies withFading and linear g

Figure: Comparison of policies withFading; g(x) = log(1 + x)

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 17 / 27

Combining Queing Theory and Information Theory

Theorem

Reliable Communication with stable data queue is possible iff

E [A] < 12 log

(1 + E [Y ]

σ2

).

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 18 / 27

MAC Policies

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 19 / 27

Information Theoretic Capacity

R1 61

2log

(1 +

E [Y (1)]

σ2

)R2 6

1

2log

(1 +

E [Y (2)]

σ2

)R1 + R2 6

1

2log

(1 +

E [Y (1) + E [Y (2)]

σ2

)Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 20 / 27

Opportunistic Scheduling for Fading Channels: OrthogonalChannels

Hk(i) = channel gain of Qi in slot k

Throughput optimal policy:

Choose queue with index

i∗k = argmax(qk(i)gi (Hk(i)(fracE [Y (i)]− εα(i))))

and use Tk =E [Y (i∗k )]−ε

α(i∗k )

α(i∗k ) = fraction of time slots assigned to i∗k estimated via LMS.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 21 / 27

Opportunistic Scheduling: CDMA

Zigbee, WIFI use CSMA.

Choose backoff timer of Qi as

f (qk(i)gi (hk(i)E [Y (i)]−εα(i) ))

f non-increasing

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 22 / 27

Figure: Orthogonal Channels:Symmetric, 3 Queues.

Figure: CSMA: Mean Delay, Symmetric10 Queues.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 23 / 27

Multihop Model

N stationary nodest sink nodesSlotted system with slot length TSensor nodes sense a random fieldDn set of sink nodes for node n. (Multicasting)A node can be in sleep or wake modeNodes generate energy via a harvesting source.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 24 / 27

Aim

Obtain a Joint Power Control Link Scheduling, Routing and Sleep-wakepolicies to maximize the throughput in a fair manner.

Approaches considered

1 APP R : Multicommodity flow model

2 APP T : Using Steiner Tree

3 APP Nc: Using Network Coding

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 25 / 27

Figure: Layout of the network : 20sensors, 3 sinks, 10 sensor nodesmulticast to sinks 1 and 2; 10 to sinks 2and 3

Figure: Performance of ALGO-M : Usedto solve OPT-R and OPT-NC

ALGO-M can provide solution for comparatively larger network.

OPT-NC provides significant improvement once OPT-R

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 26 / 27

Conclusions

Communication infrastructure has heavy cost of energy consumption,high carbon footprint.

Careful design can reduce energy and carbon footprint substantially.

Green communications requires redesign at each level.

Research Opportunities at each level.

Communication systems with energy harvesting can be designed withminimal effect of random, unreliable energy sources.

Vinod Sharma, Indian Institute of Science, ECE ()Fundamental Limits for Communication Systems with Renewable Energy SourcesApril 5, 2012 27 / 27

Recommended