Concept of Adaptive Transmission
0 likes1,082 views
The document presents a comprehensive overview of adaptive transmission techniques, focusing on optimizing data rates, power control, and channel knowledge at the transmitter. It discusses the importance of adapting transmission parameters based on channel conditions to enhance performance and reduce errors in communication systems. Additionally, the document reviews historical developments and current trends in adaptive modulation and communication strategies for various network scenarios.
![3/34
An example - cont.
10 20 30 40 50 60 70 80 90 100
−20
−10
0
10
20
Received power
Power[dB]
10 20 30 40 50 60 70 80 90 100
−20
−10
0
10
20
Power control
Power[dB]
10 20 30 40 50 60 70 80 90 100
−20
−10
0
10
20
Power control + Beamforming
sample #
Power[dB]
• Fig.: impact of beamforming and power control to the received SNR](https://image.slidesharecdn.com/pres3-151106225757-lva1-app6892/85/Concept-of-Adaptive-Transmission-3-320.jpg)
![26/34Ad-hoc versus cellular networks
B
A
A
B
• random network topology
– dynamic resource allocation and routing
– multiple hops to connect two arbitrary nodes ⇒ robust design
• WLAN, battle-field, emergency networks, medical sensors
• evolution of cellular systems
• Which one provides higher (network) capacity2
?
2
2
[Goldsmith, Cover, 1999]](https://image.slidesharecdn.com/pres3-151106225757-lva1-app6892/85/Concept-of-Adaptive-Transmission-26-320.jpg)
Concept of Adaptive Transmission
- 1. A Gentle Introduction to Adaptive Transmission Pavel Loskot University of Alberta, Edmonton, Alberta, Canada March 19, 2002
- 2. 2/34 An example • Fig.: 3-sectored single cell communication system with power control, beamforming and perhaps scalable sectors
- 3. 3/34 An example - cont. 10 20 30 40 50 60 70 80 90 100 −20 −10 0 10 20 Received power Power[dB] 10 20 30 40 50 60 70 80 90 100 −20 −10 0 10 20 Power control Power[dB] 10 20 30 40 50 60 70 80 90 100 −20 −10 0 10 20 Power control + Beamforming sample # Power[dB] • Fig.: impact of beamforming and power control to the received SNR
- 4. 4/34 An example - cont. DEM CH.EST. AGC control power ˆγ F(ˆγ) MODdata pilots ¯S SNR ≈ γ data ∆S(γ) feedback transmitter receiverchannel • a single link (point-to-point connection) • output power St(γ) = St−1(γ) ∆S(γ)
- 5. 5/34 An example - cont. 10 20 30 40 50 60 70 80 90 100 0 0.5 1 1.5 2 Transmitted Pilot Symbols 10 20 30 40 50 60 70 80 90 100 0 0.5 1 1.5 2 Channel Realization 10 20 30 40 50 60 70 80 90 100 0 0.5 1 1.5 2 sample # Received Pilot Symbols • Fig.: channel estim. via Pilot Symbol Assisted Modulation (PSAM)
- 6. 6/34 A number of questions... • What are the optimization criteria ? Is channel inversion the best ? • How often and how much to update transmit power ? St(γ) = St−1(γ) ∆S(γ) – Impact of channel delay and Doppler spread ? • How to obtain channel knowledge at the transmitter ? • How will the realistic feedback affect the performance ? • How different is the single link and multiuser case ? • Channel knowledge at the transmitter, what else can we do ?
- 7. 7/34 Conventional solution ¡ ¢ £¤ ¥ £¤ ¥ ¦ §¨ §© § § § ¦ §¨ © ! # $% ' () $% ' • Fig.: bit-error rate (BER) performance • system design for the worst case or average channel conditions ⇒ we sacrifice BER or waste power • improvements – adaptive receivers to track the channel – multiuser detection to resolve multiuser interference
- 8. 8/34 New solution a priori a posteriori ¡ ¢£ ¤¥ ¦ § §¨ ¡ © ¨ ¨ ¦ ¨ ¡ ¦ ¤¨ ¡ ¨ ¡ ¢ ¡ ¨ • Fig.: general communication system model • now the channel knowledge at transmitter is available ⇒ a new design for all channel and traffic conditions, i.e. match the transmission to the channel and traffic conditions – “match” in a sense of some optimization criterion(ia) – almost always it means to avoid bad transmit/receive conditions
- 9. 9/34 We may go further... Transmitter Receiver Noise Source Traffic Source a priori a posteriori Source • Fig.: even more general communication system model • “forward-feedback” to inform receiver about current source properties – adapt receiver algorithms – match the source to the channel • source types – ideal EP-IID data symbol source – variable bit rate, e.g. video – constant bit rate, e.g. speech – available bit rate, e.g. data transfer
- 10. 10/34Optimization problems ¡£¢ ¤¥ ¦ §¨ © © ¥ ! ¥ #%$ '( ) '01 243 05 6 ( 5 3 © 7 6 5 ¨ ) 0 8 2 ' 5 8 249 @ § A¨ • Maximize (data) rate • Minimize – power (energy), bandwidth, complexity, delay and BER – plus distortion for multimedia transmissions • Always constrained optimization – instantaneous – short-term average of instantaneous – long-term average of instantaneous
- 11. 11/34 Reliability–Integrity–Complexity Trade-off • How difficult is to approach channel capacity ? – let transmission rate R = (1 − )C and decoder probability p – let complexity χ( , p) in operations per information bit lim →0 χ( , p) , lim p→0 χ( , p) ? • Reliability – performance, robustness, BER, power efficiency • Integrity – throughput, capacity, spectral efficiency • R-I-C Trade-off cannot be avoided – design with prescribed delay, memory and computational complexity is unknown
- 12. 12/34 Feedback transmitter receiverforward channel data in feedback channel data out • the only way to get channel knowledge, a.k.a channel-state information (CSI), at the transmitter • whether negative or positive depends on the optimization problem – positive feedback – to maximize efficiency – negative feedback – to stabilize system • the same principle for TDD and FDD systems – TDD systems - channel reciprocity holds (implicit feedback) – FDD systems - explicit feedback
- 13. 13/34 Feedback - cont. • it may both simplify and complicate the system design • it allows – coordination of users – iterative solution to optimization problems – tracking of time varying channel and traffic • also internal feedbacks inside the transmitter and receiver, (e.g. turbo decoding) • it is only a “channel” (implicit or explicit), hence realistic feedback is – power and bandwidth limited – noise limited – delay limited
- 14. 14/34 Adaptation at physical layer • Source coding – unequal error protection, layer and subband coding (multimedia), • Channel coding – to cover SNR gap and/or increase SNR margin – block and convolutional codes are spectrally inefficient – coset codes (lattice and trellis) better but still “far” from capacity – turbo codes, lots of hope • Modulation – Adaptive Modulation Scheme (AMS) – multidimensional - multicarrier (MCM) and multiantenna ∗ beamforming (accurate channel knowledge) ∗ switched-diversity (moderate channel knowledge) ∗ space-time coding (no channel knowledge) – adaptive CDMA
- 15. 15/34 Adaptation at physical layer - cont. • Precoding – beamforming, switched diversity – predistorition, pre-equalization (prevent noise enhancement and simplify the receiver) – pre-Rake • Other – design of spreading codes (to match the channel) – wavelets
- 16. 16/34 Adaptation at higher layers L7 Applications Layer L6 Presentation Layer L5 Session Layer L4 Transport Layer L3 Network Layer L2 Link Layer L1 Physical Layer Medium Application Programs, Users A B • Fig.: the OSI reference model • mutual synergy among all layers • adaptive protocols and intelligence nodes • “adaptive users” (Internet)
- 17. 17/34 Adaptation at higher layers - cont. • Medium Access Protocol (MAC) and Radio Resource Management – dynamic channel allocation – variable packet length – avoid or minimize interference, packet collisions, retransmissions • Packet Reservation Multiple Access (PRMA) • ARQ – hybrid type II, incremental retransmission • routing – find the “best” route – combinatorial optimization problem
- 18. 18/34 AMS - the steps of operation • Transmission 1. Predict the channel quality 2. Choose new transmission format/parameters 3. Optional signaling of new the format/parameter • Reception 1. Estimate channel quality 2. Decide on transmission format/parameters ⇒ coordination between transmitter and receiver necessary (!)
- 19. 19/34 Step 1: Channel prediction • channel knowledge at transmitter is necessary (in our case) Hence • fully noncoherent systems do not apply • it requires bidirectional link (point-to-point connection) • prediction in order to make the system causal • typically use channel quality measures (SNR, BER) • TDD systems – open-loop adaptation – packetized transmission – faster tracking (however, channel decorrelates over time) • FDD systems – feedback link necessary – realistic channel-feedback impair the performance – better performance in presence of interference
- 20. 20/34 Step 2: Choice of new parameters • to solve given optimization problem • Data rate – constellation size, typically linear modulations (MQAM) – to vary symbol size is impractical (synchronization, bandwidth) • Power • In ergodic channels, capacity is achieved via – variable rate and power A. Goldsmith, “The capacity of downlink fading channels with variable rate and power”, IEEE Trans. Vehic. Tech., vol. 46, no. 3, pp. 569–580, Aug. 1997 – variable power (fixed rate) G. Caire, G. Tarico, and E. Biglieri, “Optimum power control over fading channels”, IEEE Trans. Inform. Th., vol. 45, no. 5, pp. 1468–1489, July 1999
- 21. 21/34Step 3: Signaling of parameters1 − Evaluate perceived channel quality − Signal the requested trans. mode − Evaluate perceived channel quality − Signal the requested trans. mode − Evaluate perceived channel quality − Decide on MS trans. mode − Evaluate perceived channel quality − Decide on BS trans. mode − Evaluate perceived channel quality − Infer the BS trans. mode blindly − Decide on MS trans. mode − Evaluate perceived channel quality − Infer the MS trans. mode blindly − Decide on BS trans. mode BSMS Uplink Downlink (b) Non−reciprocal channel, closed−loop signalling (a) Reciprocal channel, open−loop control Signal modem modes to be used by BS Signal modem modes to be used by MS BSMS Downlink Signal modem modes used by MS Signal modem modes used by BS Uplink BSMS Uplink Downlink no signalling no signalling (c) Reciprocal channel, blind modem−mode detection • note: blind decisions on modem mode increases efficiency 1 1 From: L. Hanzo, W. Webb, T. Keller, Single- and Multi-carrier QAM, 2nd ed., Wiley, 2000
- 22. 22/34 Practical considerations • closed-form solutions to optimization problem hardly exist – iterative solution, greedy algorithms, (non)linear programming etc. • AMS is affected by – finite information granularity (bits) – maximum constellation size – finite update rate – latency problem in slowly fading (large buffers) – tracking problem in fast fading • realistic feedback • channel estimation errors
- 23. 23/34 Channel knowledge • a.k.a Channel State Information (CSI) 1. nothing is known 2. fading statistics known • typically difficult to deal with 3. fade values known at receiver • coherent detection 4. fade values known at receiver and transmitter • both transmitter and receiver can be adaptive • channel knowledge at transmitter can be causal or noncausal (!)
- 24. 24/34 How beneficial is adaptive transmitter ? 1. single link - ergodic channel • Shannon capacity = maximum error-free data rate • for most distributions the gain is negligible M.-S. Alouini and A. Goldsmith, “Comparison of fading channel capacity under different csi assumptions”, in Proc. VTC, 2000, vol. 4, pp. 1844–1849. 2. single link - delay limited • capacity-versus-outage (maximum rate during non-outage) and delay-limited a.k.a zero-outage capacity (maximum rate in all fading conditions) • the less ergodic, the higher gain E. Biglieri, G. Caire, and G. Tarico, “Limiting performance of block-fading channels with multiple antennas”,IEEE Trans. Inform. Th., vol. 47, no. 4, pp. 1273–1289, May 2001 3. multiple users • gain significant for all scenarios
- 25. 25/34 AMS in networks • users mutually interfere – to optimize one link create more multiuser interference • optimization problems – orders of magnitude more complex to solve – to find global optimum is difficult – individual or joint constraints • it is not clear how to distribute control – centralized versus distributed control – note that feedback links allow coordination of users • AMS especially appealing for emerging ad-hoc networks
- 26. 26/34Ad-hoc versus cellular networks B A A B • random network topology – dynamic resource allocation and routing – multiple hops to connect two arbitrary nodes ⇒ robust design • WLAN, battle-field, emergency networks, medical sensors • evolution of cellular systems • Which one provides higher (network) capacity2 ? 2 2 [Goldsmith, Cover, 1999]
- 27. 27/34Loading algorithms Subchannel index0 1 2 3 54 6 7 Energy const transmit power channel gain • Task – find the distribution of bits and power over subchannels – conditioned on the perfect knowledge of the subchannel gains – subject to power constraints or fixed data rate • Solution – always water-filling (optimum and near optimum algorithms due to J. Cioffi3 et al. however suitable only to wireline applications) – combinatorial optimization problem, “traveling salesman” – hence it is NP-complete (=very complex) 3 3 http://www.stanford.edu/class/ee379c/
- 28. 28/34 Historical perspective - 60s 1962 R. Price, “Error probabilities for adaptive multichannel reception over of binary signals”, MIT Tech. Rep. 1963 J. C. Hancock and W. C. Lindsey, “Optimum performance of self-adaptive systems operating through Rayleigh-fading”, IEEE Trans. Comm. Syst. 1964 J. L. Holsinger, “Digital communications over fixed time-continuous channels”, MIT Tech. Rep. (water-filling in colored Gaussian channels) 1965 G. L. Turin, “Signal design for sequential detection systems with uncertainty feedback”, IEEE Trans. Inform. Th. 1966 J. P. M. Schalkwijk, “A coding scheme for additive noise channels with feedback”, IEEE Trans. Inform. Th. 1968 J. F. Hayes, “Adaptive feedback communications”, IEEE Trans. Com.
- 29. 29/34 Historical perspective - 70s, 80s 1972 J. K. Caver, “Variable-rate transmission for Rayleigh fading channels”, IEEE Trans. Com. 1973 C. E. Shannon, ”Feedback in communication problems received inadequate attention in Information Theory”, the first Shannon lecture 1972 F. E. Glave, “Communication over fading dispersive channels with feedback”, IEEE Trans. Inform. Th. 1974 R. Srinivasan and R. L. Brewster, “Feedback communications in fading channels”, IEEE Trans. Com. 1974 V. O. Hentinen, “Error performance for adaptive transmission on fading channels”, IEEE Trans. Com. • lack of good channel estimation techniques • hardware limitations 1988 J. M. Jacobsmeyer, “An adaptive modulation scheme for bandwidth-limited meteor-burst channels”, Proc. Milcom
- 30. 30/34 Historical perspective - 90s 1995 W. T. Webb and R. Steele, “Variable rate QAM for mobile radio”, IEEE Trans. Com. 1997 A. J. Goldsmith and S.-G. Chua, “Variable-rate variable-power MQAM for fading channels”, IEEE Trans. Com. 1997 A. J. Goldsmith, P. Varaiya, “Capacity of fading channels with channel side information”, IEEE Trans. Inform. Th. 1998 S. Sampei, N. Morinaga et. al, “Laboratory experimental results of an adaptive modulation for wireless multimedia communication systems”, Proc. PIMRC 1998 D. L. Goeckel, “Strongly robust adaptive signaling for time-varying channels”, Proc. ICC
- 31. 31/34 Historical perspective - 90s (cont.) 2000 G. Caire, G. Tarico, and E. Biglieri, “Optimum power control over fading channels”, IEEE Trans. Inform. Th. 2000 M.-S. Alouini and A. J. Goldsmith, “Adaptive modulation over Nakagami fading channels”, Kluwer Journal on Wireless Communications 2000 T. Keller and L. Hanzo, “Adaptive multicarrier modulation: a convenient framework for time-frequency processing in wireless communications”, IEEE Proc. 2000 L. Hanzo, W. Webb and T. Keller, “Single- and Multi-carrier Quadrature Amplitude Modulation”, Wiley, 2nd ed. 2001 S. T. Chung and A.J. Goldsmith, “Degrees of Freedom in Adaptive Modulation: A Unified View”, IEEE Trans. Com. 2001 E. Biglieri, G. Caire, and G. Tarico, “Limiting performance of block-fading channels with multiple antennas”, IEEE Trans. Inform. Th.
- 32. 32/34 Current cellular systems and standards4 • coverage 90 − 95% for certain Quality-of-service (QoS) • hence excessive SNR to support higher data rates Standard Method Method Feedback Peak of rate of format of channel data adaptation indication quality rate IS-95B code aggregation separate messages pilot strength 64 kbits/s measurements CDMA2000 variable spreading separate message pilot strength 614 kbits/s and coding blind detection measurement, power control bits W-CDMA variable spreading separate fields in Measurements: 2048 kbits/s and coding each frame •pilot strength •SINR •BER •path loss GPRS time slot separately ARQ message: 160 kbits/s aggregation and coded bits •SINR adaptive coding •av. BER •BER variance GPRS-136 time slot aggreg. separately ARQ message in 44 kbits/s AMS, incremental coded field uplink, downlink redundancy packet feedback EGPRS time slot aggreg. separately ARQ message: 474 kbits/s set of mod.+coding coded field •SINR schemes, incr. red., •BER aggresive reuse fac. •fading rate 4 4 From: S. Nanda and K. Balachandran and S. Kumar, “Adaptation techniques in wireless packet data services”, Comm. Mag., vol. 38, no. 1, Jan. 2000
- 33. 33/34 Current broadband standards • Hiperlan II – wireless ATM in 5 GHz by ETSI – physical layer - adaptive OFDM with 52 subcarriers in 20 MHz channel - variable data rate , 9, 12, 18, 27, 36 Mb/s, and optional 54 Mb/s - M-ary QAM with punctured convolutional (turbo) codes – link layer - adaptive MAC, ad-hoc capabilities • IEEE 802.11-a – also in 5 GHz band but not H/2 compatible, by IEEE WLAN standardization body – physical layer - adaptive OFDM with 52 subcarriers in 20 MHz channel - data rate 6, 12, 24 Mb/s and optional 9, 18, 36, 48 and 54 Mb/s – link layer common to all 802.11 family standards
- 34. 34/34 Conclusions Signal−to−noise ratio Performance o p e n r e s e a r c h a r e a Ergodic capacity Fixed schemes Delay−limited capacity Goldsmith et al., Biglieri et al. Biglieri et al. Goeckel Hanzo et al., Morinaga et al. • Fig.: upper and lower bound approach to adaptive modulation