evaluating dynam ic ofdma subchannel allocation for w
play

Evaluating Dynam ic OFDMA Subchannel Allocation for W ireless Mesh - PowerPoint PPT Presentation

Evaluating Dynam ic OFDMA Subchannel Allocation for W ireless Mesh Netw orks on SDRs Adrian Loch, Robin Klose, Matthias Hollick firstnam e.lastnam e@seem oo.tu-darm stadt.de Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs |


  1. Evaluating Dynam ic OFDMA Subchannel Allocation for W ireless Mesh Netw orks on SDRs Adrian Loch, Robin Klose, Matthias Hollick firstnam e.lastnam e@seem oo.tu-darm stadt.de Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch

  2. Goal & Overview Enable OFDMA PHY-layer for W ireless Mesh Netw orks  Challenge 1: how to allocate resources to nodes in a distributed mesh?  Challenge 2: how to prototype such a system on SDRs? Who can transmit? Who can receive? Who uses which OFDM subcarriers? Challenge 1 Challenge 2 Results OFDMA WARPLab limitations Exploiting good CSI Subset networks One step measurem. Avoiding bad CSI Allocation strategies Practical issues Pros and Cons Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 2

  3. Allocating Subcarriers in a Wireless Mesh Network CHALLENGE 1 Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 3

  4. OFDMA OFDMA divides bandw idth into flat-fading subcarriers  Each subcarrier can be assigned to a different link, not to a node  Gain is achieved by choosing subcarriers with good channel conditions  Requires Channel State Information (CSI) at the transmitter CH1 CH2 CH3 Amplitude Amplitude Amplitude Amplitude Subcarriers f Subcarriers f Subcarriers f Overall f A 1 B 1 A 1 B 1 A 1 B 1 ... ... ... ... A m B n A m B n Uplink (m ≠ 1, n = 1) Downlink (m = 1, n ≠ 1) Mesh (m ≠ 1, n ≠ 1) Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 4

  5. Subset netw orks Optim izing allocation for the w hole netw ork is problem atic  Signaling CSI over multiple hops may become prohibitive and too slow  Divide the network into smaller, fully-connected subsets of nodes  Optimize allocation for each subset network individually m n Step 1 Step 2 Step 3 Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 5

  6. Allocation Strategies For each subset, w e distribute N subcarriers to m x n links  First, a link is chosen using a certain strategy, which can be fair or not  Second, one or more subcarriers of the chosen link are assigned  Fair strategies assign to each link the same number of subcarriers Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 6

  7. Allocation exam ple Subchannels L4 L1 L1 L2 L4 L2 L3 L3 N/ A N/ A N/ A N/ A N/ A N/ A N/ A N/ A Links fair smin 0 0 0 0 2 2 2 2 SC 4&5 SC 1&6 SC 2&8 SC 3&7 Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 7

  8. Protocol Overview 1. Network is divided into fully connected subset networks 2. Nodes are grouped into transmitter and receiver groups Dem o! 3. CSI is measured for all m x n links in the subset network 4. CSI feedback is sent to transmitters 5. Each node calculates allocation based on CSI and strategy Dem o! 6. Data is sent according to allocation Dem o! Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 8

  9. OFDMA Prototype on WARP CHALLENGE 2 Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 9

  10. Rapid Prototyping Lim itations I m plem entation on W ARPLab for fast and easy results  Rapid prototyping allows flexible implementation, but is not real-time  Transferring data from Matlab to boards and back is slow  While suitable for lab experiments, real-world changes much faster Channel is Milliseconds measured ( ~ m s) Samples are calculated in Matlab Rapid Prototyping Cycle elapse! Samples are transferred to TX Channel allocation is board via Ethernet ( ~ m s) calculated in Matlab Channel may change much Data is sent over the actual Data is sent according to faster! wireless medium ( ~ µs) channel allocation ( ~ m s) Received signal is sent back to over-the-air Matlab via Ethernet ( ~ m s) Data is post-processed offline Switch Matlab Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 10

  11. I m plem entation Challenge Overcom e lim itation of slow CSI feedback  While still using Matlab for easy and flexible implementation …  … run allocation algorithms on up-to-date channel information  Provide a framework that can be used for any allocation algorithm Channel is Channel allocation is Data is sent according to measured ( ~ m s) calculated in Matlab channel allocation ( ~ m s) t Essentially, we want to build a time machine! Channel is Channel allocation is measured ( ~ m s) calculated in Matlab Data is sent according to channel allocation ( ~ m s) t Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 11

  12. One Step Measurem ent W e m easure CSI and perform the transm ission in one step  Key technique is to exploit coherence bandwidth b coh of the channel  b coh is range of frequencies over which channel quality is constant  Subcarriers are grouped to subchannels which are narrower than b coh 1 2 3 4 5 6 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 5 6 7 1 2 3 4 5 6 7 8 9 0 1 2 3 4 f f Large coherence bandwidth Small coherence bandwidth The performance on each subcarrier of a subchannel is similar Knowing the performance on one subcarrier, the rest can be extrapolated Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 12

  13. One Step Measurem ent Allocate one subcarrier to each sender in each subchannel  All transmissions follow such a static allocation for transmission  Each receiver gets data from each transmitter on each subchannel  Thus can deduce performance of each transmitter on each subchannel S ch 1 S ch 2 S ch 3 S ch 4 S ch 5 S ch 6 T T T T T T T T T T T T T T T T T T X X X X X X X X X X X X X X X X X X 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 There must be at least the same number of See in retrospect how any allocation subcarriers per subchannel than transmitters would have performed Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 13

  14. Practical I ssues CFO is calculated and corrected based on know n pilots  Low quality subcarriers have strong impact on average CFO estimation  Solution: weight the contribution of each subcarrier according to SNR Conventional approach: Ideal, no noise measure these distances and average W eighted approach: weighted Red symbol average, giving less importance is noisy to the noisy subcarriers CFO rotation Perform ance of allocation is m easured in CAP/ SER/ BER  CAP is the capacity based on the SNR measured on each subcarrier  Throughput cannot be directly measured due to WARPLab delays Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 14

  15. One Step Measurements in an OFDMA Testbed RESULTS Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 15

  16. Exploiting good subchannels Perform ance of strategies in a 4 x4 subset netw ork  One WARP for TX and one for RX, modeling each antenna as a node  We measure the CAP, SER and BER for all our allocation strategies  SER fair dmin is best, as overall bad links at least get best subcarriers Allocation CAP Gain Strategy [ m bps] OFDM OFDM 174.3 0% best dmax 224.3 28.7% fair dmax 192.0 10.2% fair rand 192.9 10.7% fair smin 192.7 10.5% fair dmin 192.8 10.6% Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 16

  17. Avoiding noisy subchannels W e place interferers to send noise on part of subcarriers  Setup is identical, but two of the four senders transmit artificial noise  Noise bandwidth can be increased from 0 to 5 MHz in steps of 0.5 MHz  Interference has strong impact on fair dmax , which performs as OFDM Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 17

  18. Discussion & Conclusion Discussion I ssue Advantage Lim itation Mesh Dividing network into subsets gives Forming optimal subsets in a distributed OFDMA tradeoff between local and global manner is not straightforward Allocation Provide capacity increase of 10% to Fairness limits the capacity increase and Strategies 30% and reduce BER by factor of 10 non-fair strategies may not serve nodes One Step Allows to evaluate OFDMA subcarrier Coherence bandwidth of the channel Measurem. allocations in retrospect needs to be large enough for all TXs Extensibility Other schemes requiring CSI are Schemes such as MIMO cannot use this possible, e.g., adaptive bit rates idea as # of possible signals is infinite Conclusions  We design allocation strategies for OFDMA with multiple TX and RX  We evaluate OFDMA using WARPLab and one step measurements  We achieve significant capacity and BER/ SER improvements Evaluating Dynamic OFDMA Subchannel Allocation for WMNs on SDRs | Adrian Loch 18

  19. Contact A Rapid Prototyping Framework for Practical OFDMA Systems using SDRs | Adrian Loch 19

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend