Dynamic Bandwidth Allocation of OFDMA LTE System with Game Theory

  • View

  • Download

Embed Size (px)


DPS 861A Josua Purba Jill O'Sullivan Raul Zevallos Sergio Boniche China Pankey. Dynamic Bandwidth Allocation of OFDMA LTE System with Game Theory. What is Resource Management on Cellular System ? Current Research on LTE Resource Management Research Questions So What? - PowerPoint PPT Presentation

Text of Dynamic Bandwidth Allocation of OFDMA LTE System with Game Theory

Challenges on Customer Role on Distributed Agile

Dynamic Bandwidth Allocation of OFDMA LTE System with Game TheoryDPS 861A

Josua Purba Jill O'SullivanRaul ZevallosSergio BonicheChina Pankey

1OutlineWhat is Resource Management on Cellular System ? Current Research on LTE Resource ManagementResearch QuestionsSo What?LTE Technology OverviewWhy use Game Theory?Research MethodologyFuture ResearchConclusion2 What is Resource Management on Cellular System ?What are the resources?Bandwidth (Spectrum Frequency)The RF Spectrum frequency where the signal information are sent.Limited in size.Could be 5 MHz (WCDMA), 1.4, ,5,10,20 MHz (LTE)Affect the rate and application run on the systemControl the capacity of the system to handle the users

PowerTransmit Power of Radio signalCan cause interference to other user/sector/cell if to bigDifferent system different requirements3 What is Resource Management on Cellular System ?CodeOn Code Division Multiplex technique (CDMA family, WCDMA, HSPA)Limited number of codeCould not use too many code would cause interference, thus reduce performance. This could make receiver more complex. 4 What is Resource Management on LTE System ?Resource Management on LTE System [17]The role of RRM is essentially to :Ensure that radio resources are efficiently utilizedTaking advantage of the available adaptation techniquesServe users according to their quality of service (QoS) attributes.Usually RRM handles Mobility Management (Handover) from 1 BS to another.5 What is Resource Management on Cellular System ?The mechanisms include [17]Bearer admission controlmulti-user time and frequency domain packet schedulingQoS-awareHybrid automatic repeat request (ARQ) managementlink adaptation with dynamic switching between different transmission modes. The available transmission modes include single- and dual-codeword transmissions for multi-antenna configurationsLocalized and distributed subcarrier transmission.bearer = carrier = owner6 What is Resource Management on LTE System ?BS User Plane and Control Plane Architecture [17]

7 Current Research on Resource ManagementSpectrum pooling [10],[11]Licensed Users (LU) share spectrum with Rental Users (RU)RU get the same Bandwidth size like LURU needs to detect LU before use the spectrumInterference issue from RU to LU and vice versaWorks on FDMA/TDMA and OFDM systemStudy the packet delay, throughput and the blocking probability for a spectrum pooling system by using Markov chain. [11]

Problem with spectrum pooling:Fix size for the BW, some users do not need a lot if they just use for phone callPooling all the channel and check whether it is available.Lease the BW to Rental user, not cellular8 Current Research on Resource Management?Spectrum pooling [10]

9 Current Research on LTE Resource Management?Spectrum Pooling + Random Access [13]Spectrum Pooling use Round Robin not efficientCombine it with Random Access to improve utilization radio resources and improve throughputUse Wifi for the experiment

Heterogeneous system (TV and Wireless) [14]Share (Sell) TV spectrum to service providersUse double auction game theoryOne between TV station and service providersOne between service providers and users

Problem with Spectrum pooling + Random accessIt works by modifying wifi MAC scheduler with random AccessRandom access is not organized more for ad-hoc mode. This is for wifi

For Heterogeneous system, TV broadcasters and 802.22 WRAN service providers share the spectrum:Not Cellular network but like the ideas of double auction.

10 Current Research on LTE Resource Management?Scheduling [21] Classical scheduling goals in a communication system are to maximize utilization (throughput) and to allow communication for all users (fairness).Study the fairness vs. efficiency on OFDMA scheduling. Compare various kind of game theory criteria for cooperative bargaining. Found Kalai-Smorodinsky solutions as alternative to proportional fairness (Nash solution), both offer compromise between efficiency and fairness.The problem of this research:Focus only on schedulingCan apply this to auction theory11 Current Research on LTE Resource Management?Adaptive [15]Exploits the time diversity, frequency diversity as well as multiuser diversity in the time, frequency and user domain, respectively.Adopt a two-step allocation method to reduce the scheduling complexity and meanwhile improve the scheduling performance.Allocate users into 2 dimension frequency and time domain like grids.

The key is scheduling on time, frequency and user domain.The problem with this research:Complexity is high since it involves a lot of mathMore static, since the derivation based on the certain assumption. Different condition, different assumption might give different derivation and could give different results.

12 Current Research on LTE Resource Management?Adaptive [15]

The key is scheduling on time, frequency and user domain.13 Current Research on LTE Resource Management?Optimal Solution [9] Investigate the issue of power control and subcarrier assignment in a sectorized two-cell downlink OFDMA (WIMAX) system impaired by multicell interference.Usually with practical problem, this would not have simple closed form solution.Some of available bandwidth would be reused by different base station, subject to multi cell interference.The rest of the available bandwidth would be shared in an orthogonal way between the different base stations, no multi cell interferenceThe paper provide simpler form of general solution.The key is scheduling on time, frequency and user domain.The problem with this research:Use WIMAX not LTE. They are similar but not exactly the same.More static, since the derivation based on the certain assumption. Different condition, different assumption might give different derivation and could give different results.

14 Current Research on LTE Resource Management?Cognitive Radio [8] Propose and validate a Cognitive RRM scheme in the context of LTE network segments. Use cognitive features that provide the system with knowledge which observed from past interactions with the environment. The system will be able to apply already known solutions in timely manner when identifying a problem that has been already addressed in the past. Assume: all sub carrier use the same modulation type and power level (comment: not practical) Proposed scheme can result in significant efficiency improvement in terms of performance and network adaptation.RRM: Radio Resource ManagementThe problem with this research:Use cognitive radio only, does not use game theoryDoes not involve user decision, not distributed. More centralized.15 Current Research on LTE Resource Management?Game Theory Auction Theory [20] Develop theory on allocate wireless channel with auction theorem.Consider fair competition over independent wireless fading channel.each user submits a bid according to the channel condition (assume known in the beginning time slot)Use centralized scheduler that assign time slots according to the Nash equilibrium strategy based on users average money amount.The transmitter chooses the one with the highest bid to transmit;The problem with this research:Use wireless network, not cellular networkThis is more on mathematical theory development. We can use the technique for the auction theory part. But it does not have any cognitive and scheduling aspect. 16 Research questionsWhat is the optimum way to allocate bandwidth dynamically on OFDMA LTE system with auction theory, scheduling and cognitive radio? Is it possible to find general optimum solution?What is the complexity of the dynamic bandwidth allocation with auction theorem compare to results without game theory?How to apply time notion as multiple step decision of auction theory on allocation the bandwidth dynamically?17So What ?RIM CEO mention the need to conserve bandwidth(http://www.mobilecrunch.com/2010/02/16/rim-ceo-pulls-an-att-we-need-to-conserve-bandwidth/)At the end 2009, AT&T ask its customer to reduce to use their smart phone by giving incentive.http://news.cnet.com/8301-30686_3-10412804-266.htmlOperator can increase the capacity and efficiency of the network. Thus increase the revenue bottom line make money and customer satisfactionWhy LTE ?People/customer use the technology not only research but also commercial (real implementation)Majority market use LTE compare to Wimax and Ultra Mobile Broadband (UMB)18 Lte Overview key featuresSupport for, and mobility between, Multiple heterogeneous systems: legacy system (GSM, GPRS, EDGE, WCDMA, HSPA)Non-3GPP system (Wifi, Wimax, EV-DO, satellite)All IP Network Enhanced Air Interface allow increased data rate With Mobility: 100 MBps (DL) and 50 MBps(UL)Stationary: 1GBps (DL) and 500 MBps (UL)19 Lte Overview key featuresSupport for higher throughput and lower latencyUser Plane Latency: < 5msControl Plane Latency (Transition Time to Active State): < 100ms (from idle to active)Increase Control Plane Capacity: > 200 users per cell (for 5MHz Spectrum)Mobility Support:Up to 500 Kmph Optimized for low speed from 0 to 15 Kmph

20 Lte Overview key featuresSpectrum Flexibility to achieve higher spectrum efficiency [18]: where RB: Resource Block

21 Lte Overview key featuresChannel Bandwidth Definition [18]:

22 LTE Technology OverviewHigh Level Overview BS Architecture [19]

23 LTE Technology OverviewLTE scheduler on protocol stack [16]

24 LTE Technology OverviewChannel quality variations in time and freq [16]

25LTE Technology OverviewDown Link (DL) OFDM (Orthogonal Frequency Division Multiplexing) use a large number of narrowband sub-carrier for multi carrier transmission. OFDM avoids the problem with multipath reflections by sending message bits slow enough so it has high tolerance for multipath delay spread.OFDMA: assigning different su