Post on 19-Aug-2020
Random Packet-CDMA:Reducing Delay and IncreasingThroughput of WLAN Systems
Roland Kempter and Behrouz Farhang-Boroujeny
kempter@ece.utah.edu
farhang@ece.utah.edu
Organization
1) Introduction:Scheduled vs. Random Channel Access
2) Random Packet-CDMA (RP-CDMA)
3) RP-CDMA: Header Detection
4) RP-CDMA: Payload Detection
5) Conclusions
Introduction: scheduling-based access
Packet in queue
Transmitter (TX) Handshake w/ Base
Ask for resources
Base
Thinking about it!
TXHandshake
w/ TX
Tell about resources
AdvantagesBase knows everything:• packet collisions can be avoided• service can be guaranteed (great for voice)
DisadvantagesResources need to be negotiated:• OK in circuit switched systems• large overhead (or low efficiency) with packetized traffic
Data (IP) trafficis bursty!
Introduction: random channel access
Packet in queue
Transmitter (TX) TX
Check medium(sometimes)
TX Base
Send Acknowledgement
Advantages• self regulating (good for bursty traffic)• load adaptive• simplifies infrastructure
Disadvantages• Collisions, low stable throughput• degrades heavily as load increases • delay variations
Introduction: background on random channel access
1970
Aloha1976Ethernet
1980
IEEE 802.3CSMA
1990
Spread Aloha
1997
IEEE 802.11
2000+CDMA:ConcurrentTransmissions
Advances in (Random) Channel Access
RP-CDMA: combination of SpreadAloha with CDMA technology
Max. throughput of Spread Aloha
(18% of a fully coordinated system)
No collisions
Know SS for detection→ handshakingNot Random Protocol
S = G
RP-CDMA packet structure [KemAmiFar06, SchKemKot06]
[KemAmiFar06] R. Kempter, P. Amini and B. Farhang-Boroujeny, “Throughput and Stability of RP-CDMA andSpread Aloha in Multipacket Capture Channels", submitted to IEEE/ACM Transactions onNetworking, October 2006
[SchKemKot06] C. Schlegel, R. Kempter and P. Kota, “A Novel Random Wireless Packet Multiple Access MethodUsing CDMA”, IEEE Trans. Wireless Communications, p. 1362-1370, Vol. 5, No. 6, June 2006
Header Channel: low-traffic Spread ALOHA channelData Channel: CDMA, possibly large number of TX
System Characteristics
RP-CDMA Performance: interference in the header channelSystem EffectPacket format separates TX into a virtual header and virtual data channel
If header detection limited by (h/d) interference:→ CSMA/CA on the header unlikely to improve performance
Note
Lh<Ld and concurrent TX: feature of RP-CDMA→ header detection likely (h/d) interference limited
RP-CDMA Performance: collisions in the header channel
Time
Collision at the base station? - a packet level point of view
Payload portion
with random
spreading
Header portion
with common
spreading
Packet from node x
Packet from node y
Collisions: not necessarily
As long as no overlap at chip level: still separated by Spreading Gain N.
Collision? -
on the chip level
RP-CDMA Header Detection: results
• Header detected successfully iff after receiver, SNIR > header detection threshold
• No chip-level collisions happened• Equal power scenario
Assume
• Solve:
• Check for chip-level collisions
Matched Filter [TseHan99]
Packet timing established during header recovery:Matched Filter for header detection
RP-CDMA: header detection, spread aloha, equal powers
(b) Nh=Nd=20(a) SNRh/SNRd=5 dB
(c) Ld / Lh=25
RP-CDMA: performance
Conclusions from the header process• After a certain point which is f(N,SNR), increasing Ld / Lh
does not improve performance
Interference limitation
Collision limitation
Small systems Interference can be suppressed
Large systems Interference can onlybe suppressed up to a certain load
Pointf (Nh, Nd, K, Ph/Pd, pathloss, γ, header technology, data technology)
RP-CDMA: data channel performance
Data frame is detected successfully iff after receiver, SNIR Γ > threshold γ.
SNIR Γ is multi-user receiver specific
Assume
RP-CDMA Data Detection
• Matched Filter [TseHan99]• Decorrelator [TseHan99]
• MMSE [TseHan99]
• Partitioned Spreading [KemSch05, ShiSchKem06]: iterativedemodulation, resolves (virtually) all multiuser interference
[KemSch05] R. Kempter and C. Schlegel, “Packet Random Access in CDMA Radio Networks”,in Proceedings of Allerton Conference 2005.
[ShiSchKem06] Z. Shi, C. Schlegel and R. Kempter, “On the Performance of Partitioned-Spreading CDMA”, Conference on Information Sciences and Systems, CISS 2006, Princeton, 2006
RP-CDMA: network simulations
IP packet sizes are trimodally distributed [Inet2]:• Pr(L=50 bytes) = 0.5• Pr(L=500 bytes) = 0.4• Pr(L=1500 bytes) = 0.1
Also• RP-CDMA header size: Lh= 50 bits → E[Ld / Lh] = 60• Payload and header spreading gains: N=32• Header/Payload SNR: 5 dB• Header and data detection thresholds: 3 dB• Power control• Nodes transmit to a central base station
RP-CDMA: equal power, central base station
RP-CDMA receiver: SOFTWARE DEFINED RADIO
Laid out formax. # users
Software processes,payload decoding technologycan be adopted f(load)
Conclusions 1/2
• We demonstrated the performance in the RP-CDMAheader channel by modeling it as a Spread Aloha packet.
• After a point, RP-CDMA becomes interference rather thancollision limited:
f (Nh, Nd, K, Ph/Pd, pathloss, γ, h. technology, d. technology)
• For the data channel, we compared:- Matched Filter - Decorrelator - MMSE- Partitioned Spreading
in base station centric networks w/ power control
Conclusions 2/2
• RP-CDMA is determined by header process:improving header directly is hard (non-linear in parameters):
BUT: with Partitioned Spreading, for a targeted data rate, Pd can be low, increasing Ph / Pd→ pushing the interference limit
• We presented the general block-diagram of an RP-CDMAsoftware defined receiver
• RP-CDMA allows to adopt to different load situations viasoftware changeslow load: power save w/ matched filterhigh load: partitioned spreading
The EndAdditional References
[TseHan99] Tse, David N. C. and Hanly, S. V.,”Linear Multiuser Receivers: Effective Interference, Effective bandwidth and User Capacity”, IEEE Transactions on Information Theory, Vol. 45,No. 2, pp. 641-657, March 1999
[Loy62] Loynes, R. M, “The Stability of a Queue with Non-independent Inter-arrivals and ServiceTimes”, Proc. Camb. Phil. Soc., Vol 58, pp. 497-520, 1962
[LuoEph06] Jie Luo and Ephremides, A., “On the throughput, capacity, and stability regions of randommultiple access”, IEEE Trans. Inform. Theory, Vol. 52, No. 6, pp. 2593-2607, July 2006
[XuBae02] Xu, K. and Sang Bae, M., “How Effective is the IEEE 802.11 RTS/CTS Handshake in Ad HocNetworks?”, GLOBECOM 2002, Vol. 1, pp. 72-76, Nov. 17-21
[RayCarStar03] Ray, S. and Carruthers, J. B and Starobinski, D., “RTS/CTS-Induced Congestion in Ad hocWireless LANs”, Wireless Communications and Networking, WCNC 2003, Vol. 3, pp. 1516-1521,March 16-20
[HaaDen02] Haas, Z. J. and Deng, J., “Dual busy tone multiple access (DBTMA)-a multiple access controlscheme for ad hoc networks”, IEEE Trans. Commun., Vol. 50, No. 6, pp. 975-985, 2002
[Inet2] Statistics for the Abilene backbone network of Internet2, 2005,http://netflow.internet2.edu/
[Bia00] Bianchi, G., “Performance Analysis of the IEEE 802.11 Distributed Coordination Function”,IEEE Journal on Selected Areas in Communications, Vol. 18, No. 3, pp. 535-547, March 2000
[NiTschSh99] Ni, S. Y. and Tseng, Y. C. and Chen, Y. S. and Sheu, J. P., “The Broadcast Storm Problem in aMobile Ad Hoc Network”, ACM MOBICOMM ’99, August 1999
[GupKum01] P. Gupta and R. Gray and P. Kumar, “An Experimental Scaling Law for Ad hoc Networks”,Univ. of Illinois at Urbana-Champaign, Tech. Report, May 2001
[FLUX] David Johnson, Tim Stack, Russ Fish, Dan Flickinger, Rob Ricci, Jay Lepreau, “TrueMobile: AMobile Robotic Wireless and Sensor Network Testbed,” University of Utah Flux Group Technical Note 2005-02, April 2005. Revised version to appear in INFOCOMM 2006
[HaePuc05] Martin Haenggi, Daniele Puccinelli, “Routing in Ad Hoc Networks: A Case for Long Hops,”IEEE Communications Magazine, October 2005