microwave capacity%5f understanding techniques to improve throughput
DESCRIPTION
paperTRANSCRIPT
1
UNDERSTANDING TECHNIQUES TO IMPROVE THROUGHPUT
AVIAT TECHNOLOGY SERIES
MICROWAVE CAPACITY
“I canna change the laws o’ physics captain”
Strategies for Increasing Microwave Capacities
1. More Spectrum
(More Hz)
2. More Spectral Efficiency
(More Bits per Hz)
3. More “Effective” Throughput
(More Data per Bit)
Technique
Multiple channels with link aggregation (incl CCDP)
ACM
Technique
Higher modulations (512/1024 QAM, ACM)
FEC redundancy
Technique
Header optimization / suppression
Payload compression
Asymmetrical RF
NOVEMBER 2011 AVIAT NETWORKS | 3
Poll Question: Your Understanding of Microwave Capacity?
1. High. I understand the concepts and how to apply them
2. Medium. I understand the concepts, unclear how to apply
to my network
3. Low. I lack some understanding on concepts/technologies
4. Very low. I am completely confused
NOVEMBER 2011 4 AVIAT NETWORKS |
Focus =
RF Link
Aggregation
Multi-Channel Requires Link Aggregation: 2 Approaches
NOVEMBER 2011 5 AVIAT NETWORKS |
Ethernet Links Radio Links or Polarizations (CCDP)
LAG 802.1ax
Layer 1 Link Aggregation (L1LA)
Designed
for this
Supports
this
Custom solution
for RF links
Radio/RF Link: Aggregation Options
NOVEMBER 2011 6 AVIAT NETWORKS |
LAG 802.1ax L1LA
Approach Packet-by-Packet Byte-by-Byte
Number of links Yes (up to 2 RF links)
Yes (up to 4 RF links)
Load balancing Effectiveness
Medium High
Easy capacity expansion Yes Yes
Latency High Low
Adaptive to RF No Yes
Radio Links
Link Aggregation Recommendations and Conclusions
1. To enable N+0 radio links (which is the most important
problem to solve since radio capacity is constrained), L1LA
is the ideal solution
2. Only use LAG on Ethernet links for > 2 Ethernet ports
(higher layer protection and load balancing schemes work
best)
NOVEMBER 2011 7 AVIAT NETWORKS |
Higher Order Modulation Schemes – 1024QAM
• Improves bits/Hz efficiency within the same channel size
• 24% throughput improvement over 256 QAM
• Disadvantages in fixed operation:
• Lower system gain (Tx power and Rx threshold), leading to shorter hops or larger antennas
• Increased sensitivity to interference can reduce link density
• Increased phase noise and linearity requirements increases radio design complexity and cost
• Should be deployed in conjunction with ACM
• First commercially available systems on the market in 12-18 months
NOVEMBER 2011 8 AVIAT NETWORKS |
FEC Redundancy
• Forward Error Correction is employed as standard on all digital
microwave systems
• Enables virtually error free performance right up to Rx threshold
• Involves adding additional overhead bits (~10%), to enable bit errors to
be detected and corrected at the receiver
• FEC can be:
• Stronger – more bits enables more errors corrected – better system gain, but at the
expense of lower throughput
• Weaker – less bits enables higher throughput, but poorer system (Rx) performance
• Actual implementation can vary. Some vendors (eg: Aviat) offer both
NOVEMBER 2011 9 AVIAT NETWORKS |
ACM Capacity Improvement: Example 1
• Requirement:
• Capacity: 150 Mbps
• Availability: 99.999%
• Path Length: 15.3 miles
• ACM Benefits
• 23 Mbps more capacity
for most of time
• 15% more data sent
• ~4 minutes more
uptime per year
• Same antenna, same
frequency, same
channel, etc
10 NOVEMBER 2011 AVIAT NETWORKS |
ACM Design (6 GHz, 6' Antennas)
Uptime Time in Modulation (min)
Capacity (Mbps)
Total Bytes Per Year (TB)
256QAM 99.9975% 525,587 178 701.66
64QAM 99.9991% 9 144 0.01
16QAM 99.9995% 2 94 0.00
QPSK 99.9999% 2 46 0.00
out of service 1 0
Total 525,600 701.67
Fixed Modulation Design (6 GHz, 6' Antennas)
Uptime Time in Modulation (min)
Capacity (Mbps)
Total Bytes Per Year (TB)
128QAM 99.9990% 525,595 155 611
out of service 5 0
Total 525,600 611
ACM Capacity Improvement: Example 2
• Requirement:
• Capacity: 50 Mbps
• Availability: 99.999%
• Path Length: 10.15 miles
• ACM Benefits
• 213 Mbps more
capacity for most of time
• 500% more data sent
• 28 sec more uptime per
year
• Smaller antenna =
$4,800 per year cost
savings
11 NOVEMBER 2011 AVIAT NETWORKS |
ACM Design (18 GHz, 2' Antenna)
Uptime Time in Modulation (min)
Capacity (Mbps)
Total Bytes Per Year (TB)
256QAM 99.5613% 523,294 263 1,032
64QAM 99.7784% 1,141 155 1.33
16QAM 99.9299% 796 100 0.60
QPSK 99.9995% 366 50 0.14
out of service 3 0
Total 525,600 1,034
Fixed Modulation Design (11 GHz, 4' Antennas)
Uptime Time in Modulation (min)
Capacity (Mbps)
Total Bytes Per Year (TB)
64QAM 99.9994% 525,597 50 181
out of service 3 0
Total 525,600 181
ACM gives more capacity, at lower cost
Poll Question: What’s stopping you from deploying ACM?
1. Uncertain value, not convinced of benefits
2. Unclear of FCC’s position
3. Unclear how to design/implement it
4. Existing equipment lacks ACM support
5. Nothing, already deploying/deployed
NOVEMBER 2011 12 AVIAT NETWORKS |
Ethernet Traffic Optimization
NOVEMBER 2011 13 AVIAT NETWORKS |
• Inter-frame Gap and
Preamble Removal
• 20 bytes at the
beginning of each
frame
• Payload Header
Compression
• Removes known or
restorable information
from MAC and/or IP
header
• Payload Compression
• Identifies repetitive
patterns in the payload
and compresses them
Benefit of Header Compression
NOVEMBER 2011 14 AVIAT NETWORKS |
Frame Size
Standard Frame
IFG & Preamble IFG & Preamble &
MAC header
Frame Space
Mbps Frame Space
Mbps Increase Frame Space
Mbps Increase
64 84 76.2 68 94.1 24% 58 110.3 45%
128 148 86.5 132 97.0 12% 122 104.9 21%
260 280 92.9 264 98.5 6% 254 102.4 10%
512 522 96.2 516 99.2 3% 506 101.2 5%
1518 1538 98.7 1522 99.7 1% 1512 100.4 2%
Payload Compression
• Employing common compression techniques to replace
strings of repeated patterns of data
• Pros
• Promises dramatic throughput improvement (2.5x), with minimal
additional cost or spectrum requirements
• Cons
• Improvement is not guaranteed nor predictable, since it is highly
dependent on the traffic mix
• Compression works better for uncompressed VoIP and partially filled
TDM
• No benefit for traffic that is already compressed
• Typical real-world improvement is minimal (~4%)
• Other potential issues with compression, eg: increased link latency
• Conclusion – don’t expect to see payload compression
adopted widely in the industry
NOVEMBER 2011 15 AVIAT NETWORKS |
Asymmetric Link Operation
• Involves the concept of ‘spectrum borrowing’
• Moving spectrum from the uplink of an adjacent link to increase downlink spectrum bandwidth
• Advantages
• Useful for networks with asymmetrical traffic flows
• Disadvantages
• FCC Common Carrier bands are symmetric only
• Introducing asymmetrical operation into existing bands will be almost impossible
• Requires complicated and lengthy regulatory approval
NOVEMBER 2011 16 AVIAT NETWORKS |
NOVEMBER 2011 17 AVIAT NETWORKS |
f1 f2
DL
f1’ f2’ UL f3 f4 UL
f3’ f4’
DL
Before - Symmetric Microwave Network
2x10MHz = 20MHz channel
NOVEMBER 2011 18 AVIAT NETWORKS |
f1 f2
DL
f1’ f2’ UL f3 f4 UL
f2’ f3’
DL
Asymmetric Microwave Network, after Spectrum Borrowing
3x10MHz = 30MHz channel f4’ f3
Congratulations….
you’ve increased your
microwave capacity
• You can now meet
capacity demands of
your end applications
• You can now
implement lower cost
network designs
NOVEMBER 2011 19 AVIAT NETWORKS |
Obvious
Not so obvious
Capacity Comparison by Architecture (no ACM)
NOVEMBER 2011 20 AVIAT NETWORKS |
Hub – Spoke Ring
Sites 6 6
Links 5 (1+1) 6
Antennas 10 12
Average Antenna Size 5.2 2.3
Reliability (per link) 99.999% 99.95% (average)
Capacity (per link) 50Mbps 200Mbps
CAPEX $176k $119k
Yearly Tower Lease $77k $48k
5 Year TCO $676k $478k
How do some vendors inflate throughput?
• Present throughput figures based upon 64 byte
frame sizes only
• Assume that up to 100% of payload traffic is
compressible
• Assume availability of wide channels (80 MHz)
• Assume co-channel operation (2x RF carriers)
on the same frequency assignment (using XPIC)
• Present half-duplex throughput figures
• Assume gains from other unproven techniques
NOVEMBER 2011 21 AVIAT NETWORKS |
When it comes to
Microwave Capacity
To be sure, use an industry standard benchmarking test, like RFC 2544
Best Case Throughput – 80 MHz channel
NOVEMBER 2011 22 AVIAT NETWORKS |
340
Airlink Strong
FEC
360
IFG+PA
Suppression
450
MAC HC
520
2+0
XPIC
1040
Payload
Compression
2000
1024QAM
2500
360 360
720 720* 900
‘Guaranteed’ throughput
Maximum ‘Best Efforts’ throughput 64 byte frame size, ideal traffic profile
Throughput figures are stated in Mbit/s and are approximate for a
single 80MHz RF channel and 256QAM (unless otherwise stated)
* + Latency
Realistic Throughput – 30 MHz channel
NOVEMBER 2011 23 AVIAT NETWORKS |
180
Airlink Strong
FEC
190
IFG+PA
Suppression
201
MAC HC
209
2+0
XPIC
418
Payload
Compression
435
1024QAM
544
190 190
380 380*
475
‘Guaranteed’ throughput
Maximum throughput For 260 bytes average frame sizes, and
typical traffic profile
Throughput figures are stated in Mbit/s and are approximate for a
single 30MHz RF channel and 256QAM (unless otherwise stated)
* + Latency
+4%
+6% +4%
+25%
Microwave Capacity Improvements – Hype and Availability
NOVEMBER 2011 24 AVIAT NETWORKS |
Hype Factor Availability
Multi-Carrier Low Now
Higher Modulation Medium 12-18 months
Strong FEC Low Now
ACM Low Now
Traffic Optimization Medium Now
Payload Compression High Now
Asymmetrical Operation High >3 years
Ring Networks Low Now
Aviat Advanced Microwave Technology Seminar
Email:
November 8, 9 2011
Aviat HQ
Santa Clara, CA
Day 1 Network migration - TDM to IP Carrier Ethernet Transport & MPLS LTE requirements on backhaul Ethernet radio capacity analysis Network Timing and Synchronization
Day 2 ACM Microwave Strong Security Microwave antenna tech update Outsourced network operations
NOVEMBER 2011 AVIAT NETWORKS | 25
Poll Question: Next Webinar Topic
1. Microwave Security
2. Ethernet OAM
3. Ethernet Protection and Redundancy
4. Q&A with Dick Laine (no slides)
5. Microwave Q&A (no slides)
NOVEMBER 2011 26 AVIAT NETWORKS |
THANK YOU