Slide 1

Next Generation Wireless Networks: Research Challenges and OpportunitiesAbhay KarandikarProfessorDepartment of Electrical EngineeringIndian Institute of Technology Bombay, Mumbai 400076karandi@ee.iitb.ac.in1

1OutlineGlobal Wireless ScenarioIndian ScenarioKey Research Issues and Technology Components of 4GOur Contributions to 4GBandwidth Request Procedure in IEEE 802.16mEnergy Efficient Scheduling over Wireless ChannelsGoing Forward: Towards Bridging the Digital Divide22OutlineGlobal Wireless ScenarioIndian ScenarioKey Research Issues and Technology Components of 4GOur Contributions to 4GBandwidth Request Procedure in IEEE 802.16mEnergy Efficient Scheduling over Wireless ChannelsGoing Forward: Towards Bridging the Digital Divide33Drivers for Broadband4

Plethora of services causing Internet traffic to grow 50% every year4Wireless Broadband5GSM / EDGEHSDPAWCDMA / EVDOHSPA (3G)ADSLADSL2+LTE / EVDO-Rev C (UMB)IMT-A (4G)VDSLGDONEthernetWiMAX10x Kbps1x Mbps100x MbpsData RateFixedMobile5Timelines for IMT-Advanced (4G)6

2008

2009

2010

2011

2012

2013

Proposals

Evaluation

Consensus

Specification

Early Deployment6Global ScenarioNations are becoming increasingly aware of IPR and use their IPR reserves to position trade policyChina good exampleHas pushed its IPR in major 3G/4G standardsBalance of IPR is shifting40% essential patents from AsiaOutlineGlobal Wireless ScenarioIndian ScenarioKey Research Issues and Technology Components of 4GOur Contributions to 4GBandwidth Request Procedure in IEEE 802.16mEnergy Efficient Scheduling over Wireless ChannelsGoing Forward: Towards Bridging the Digital Divide88iPhone Example9

700+ Million Cellphone subscribersSecond largest telecom market in the world Fewer products made in India large part is still imported Second largest import after OilHigh outflow of Foreign CurrencySeveral $$ outflow per handsetOnly country in the top telecom markets with no umbrella body focusing on standardsLow Geographical CoverageOnly 60% of India25% of villages coveredVery Low Broadband penetration

10Scenario in IndiaPopulation Density Map(source: www.reliefweb.int)

10OutlineGlobal Wireless ScenarioIndian ScenarioKey Research Issues and Technology Components of 4GOur Contributions to 4GBandwidth Request Procedure in IEEE 802.16mEnergy Efficient Scheduling over Wireless ChannelsGoing Forward: Towards Bridging the Digital Divide1111Technology Components for 4GOFDMA

Multi-Antenna (MIMO)

Spectrum flexibility and Carrier aggregation

12Time231122333Frequency23DiversityBeamforming

BS

BS

Aggregated carriers = 40 MHz20 MHz20 MHz12Technology Components for 4GSelf organization and Self optimization

Relaying

Enhanced Quality of Service support

13

access linkrelay linkaccess linkBSRS

13OutlineGlobal Wireless ScenarioIndian ScenarioKey Research Issues and Technology Components of 4GOur Contributions to 4GBandwidth Request Procedure in IEEE 802.16mEnergy Efficient Scheduling over Wireless ChannelsGoing Forward: Towards Bridging the Digital Divide1414Wireless Uplink Scheduling

BS computes non-conflicting schedule15

Bandwidth Request (BR) Procedure16

IEEE 802.16mAdvanced Mobile Station (AMS)IEEE 802.16mAdvanced Base Station (ABS)Gold AMSSilver AMSBronze AMSBR OpportunitiesContentionWindowContention Window DoubledAMS waits for a random number of BR Opportunities Number of BR opportunities to wait chosen from Contention WindowCommon channel for all AMSTimeBR collisionGrantGrantCurrentTimeCurrent Time16Drawbacks of BR ProcedureNo differentiationNo differentiation in window size based on service class

Lacks fairnessNew users can get channel before old users

Scaling factor of 2 is used independent of System loadNumber of retries

1717Current TimeConnection Priority (CP) for each contending Service FlowCP is a function of Service Class & Number of Retries (BR collisions) CP parameters:Service ClassInitial / Maximum Window SizeWindow Scaling Factor

Our BR Proposal (IEEE C80216m-09_1321r4)

18

IEEE 802.16mAdvanced Mobile Station (AMS)IEEE 802.16mAdvanced Base Station (ABS)Gold AMSSilver AMSBronze AMSContentionWindowContention WindowChanged as per CPTimeBR collisionGrantCurrentTime18OutlineGlobal Wireless ScenarioIndian ScenarioKey Research Issues and Technology Components of 4GOur Contributions to 4GBandwidth Request Procedure in IEEE 802.16mEnergy Efficient Scheduling over Wireless ChannelsGoing Forward: Towards Bridging the Digital Divide1919Wireless Channel CharacteristicsWireless Channel is characterized bySignal strength variation over time, frequency and spaceSmall scale variation (Fading)InterferenceLimited battery life at hosts

Physical Layer no longer a fixed rate bit pipe

Resource allocation needs to take channel characteristics into account20Significant performance gains in wireless networks by Cross-Layer Design20SNR Fluctuations in a Multiuser System21User 2User 3User 3User 1SNRTime21Multiuser Diversity: A New Form of DiversityChannel fades independently for each user so Different users experience different channel gainsHigh probability that some user will have a strong channelBS schedules the user with the strongest (best) channelHence Opportunistic SchedulingTransmitting in favorable channel condition also minimizes power but at the expense of delayScheduling- Power is minimized subject to delay constraint2222Energy EfficiencyRate-Power relationship is convex23

23Energy Efficient SchedulingSingle Receiver (Base Station) and multiple transmittersBase station is the centralized scheduler24

24Energy Efficient SchedulingQueue transition, average queue length, average power for user i

Problem: Minimize the power consumption of each user subject to delay constraint of each user

Multi-objective constrained optimization problem

25

25Uplink SolutionVisualize a link between user and base station as a Point-to-Point scenarioEach userDetermines its transmission rate as if it was the only userInforms this rate to the base stationThe base station schedules the user with the highest rateQueue transitions for a user who is scheduled and not for othersPower and queue cost are appropriately updated26

26Uplink Solution : Auction InterpretationThe base station auctions each time slotThe user quoting the highest rate wins the bidUser quote rates that are just sufficient to satisfy their delay constraintsQuoting unnecessarily high rates not favorable since power minimization is the objectiveIt can be proved that the queue lengths converge to cooperative equilibrium and delay constraints are satisfied

27Nitin Salodkar, Abhay Karandikar, Vivek Borkar, A Stable On-line algorithm for Energy Efficient Scheduling IEEE Transactions on Mobile Computing October 201027OutlineGlobal Wireless ScenarioIndian ScenarioKey Research Issues and Technology Components of 4GOur Contributions to 4GBandwidth Request Procedure in IEEE 802.16mEnergy Efficient Scheduling over Wireless ChannelsGoing Forward: Towards Bridging the Digital Divide2828Driving Research Agenda29ChallengesBackhaul connectivityLow cost wireless backhaulLow ARPULow cost infrastructure Tight integration with IPInfrastructure sharingDSL like experienceArchitecture for high speedHigh speed spectral efficiency at cell edgePower SupplyVery high energy efficiency29

Present ScenarioBackhaul

PSTNTDM/ ATM/IP / MPLS

EdgeRouterInternet Voice VideoAudioPacket switching Node Gateway

Access & DevicesCore Services Backhaul

TDM / ATM/ETHERNETMSC

GSM /EDGE

UMTS/ HSPA

WiFi

BSC/RNC Data centre

30

DSLAM

Optimized MAC for outdoor point to point linkInterference management for multi-hop linkE1 circuit emulation for Legacy access GSM/CDMA)Inexpensive radio (