Tuesday, February 24, 2015

Are spot beams better than wide beams for aviation?



Are spot beams better?
The marketing of spot beams suggests that the more powerful transponders will improve the link budget and overall economics over legacy wide beam transponders.  Costs are driven by commercial factors that scale the results based on limited public information.  Technical performance can be modeled quite precisely, but whose accuracy is subject to assumptions not always readily verifiable to a public reference.

Using my best judgement, the following analysis dives deep into each aspect to reveal the key differences and benefits of Ku band spot beams.

Will it cost less using spot beams?

How much cheaper?   First let's use precise definitions and here's the fine print.

What is the cost per hour (assuming 10 hours of use per day) for a 1 Mbps streaming service (combined forward and return by varying ratios)?
NOTE: this Ku-band analysis is purposely simplified to allow ready comparison to various scenarios. The actual performance of a satcom system may be substantially different.  
Three different streaming ratios bracket the service levels:
  1. 4:1   (1 Mbps forward and 256 kbps return) - Internet Access 
  2. 8:1   (1 Mbps forward and 128 kbps return) - Internet Access
  3. 16:1 (1 Mbps forward and 64 kbps return) - Movie streaming 480p 30 fps

Service was modeled at 10 hours a day.  Higher utilization reduces costs proportionally.
Two different fabricated antennas in two different steering solutions.  
  1. 34x7 typical mechanically steerable along two gimbals
  2. 30x30 antennas steerable electronically or along one gimbal 
  1. 30 deg elevation and 30 deg skew angle for "disadvantaged" scenario
  2. 60 deg elevation and 10 deg skew angle for "good" scenario

Spot beam and wide beam proprietary link budget models 
Five contours to represent beam edge to beam peak 
Fabricated modem assuming adaptive forward and return modcods, and clear sky 
Ka band analysis is not included. It should be noted that Ka-band economics are structured differently, with the satellite operator providing the service in Ka band Versus leasing commercial transponders in Ku band. 
See Link Budget notes below for more detail.

The following chart shows the baseline scenario.





What if the price per MHz were $2000, regardless legacy wide beam or spot beam?

The following chart shows the same scenario as above, except both the spot beam and wide beam spectrum is priced at $2000 per MHz.






On equal footing with spectrum pricing, the 30x30 legacy wide beam cost is less than with spot beam.   The 34x7 costs were similar spot beam to wide beam.



Spot Beam Uplink Interference

A spot beam transponder uplink is like a dog's breakfast...

The performance of the spot beam service is impacted severely by uplink interference.  

I have judged the level of interference in my model based on publicly released link budgets from more than one service provider which shows surprisingly severe interference.  Not being certain if the effect is typical or exceptional, I have also looked at a second case assuming spot beam uplink interference is reduced from the baseline (but still worse than legacy wide beam.) This remains an area of investigation - if new info emerges I will revise the results as applicable in this post.

In the chart below, the baseline case is repeated with a reduction in uplink interference:





In all cases, the 30x30 antenna offered substantial savings in session cost over the 34x7 antenna.

A better antenna pays dividends regardless of the beam.


Costs are driven almost directly by the basic cost per MHz, regardless of the beam.

Uplink interference is limiting spot beam spectral efficiency with both forward and return channels.



Cost of Streaming
With the baseline $2000/$3500 cost per MHz differential, and accounting for varying interference, then Spot beam 16:1 1 Mbps service costs are estimated between $9-$16 for the 30x30 antenna and $12-$26 for the 34x7 antenna

A retail price point of $10 per hour for personal streaming seems to be gaining traction.

30x30 1 Mbps (16:1) streaming session costs are $9-$16 per hour.

For a streaming session to be profitable:

  • the average streaming rate should be closer to 500 kbps, which reflects either lower image quality
  • a "breakage model" where the time charges accrue even while the streaming session is not active - for example while a session is paused.
  • higher than 10 hour per day utilization, perhaps through complementary services
  • lower than $2000 per MHz pricing could bring cost per session below $10 an hour.
Cost of Internet Access
A 1 Mbps Internet access session (assume 4:1) can accrue over 600 MB in one hour.  Assuming the $2000/$3500 (Spot/Wide) price per MHz and accounting for varying interference, spot beam cost is about $10-17 for one hour whereas $20-$35 with wide beam (comparing 30x30 Vs 34x7 antennas). 

Spot beam costs are one-half the cost of wide beam, because the spectrum cost is so much less.

A cost of $10-35 an hour equates to about $0.017 - $0.059 per MB.  These costs assume the channel is occupied constructively all the time.

30x30 100 MB Internet Access (4:1) would cost $1.70 - $2.80


Will spot beam be faster and more efficient?

With the extra uplink interference in spot beams coupled with a hard downlink PSD limit, spot beams perform poorer than legacy wide beams.  There is a silver lining (wider bandwidth).

To explore each scenario, five solutions are presented to represent performance from beam edge (on the left) to beam peak (on the right).

Forward Channel
The following chart shows the difference in data rate on the forward channel (spot - wide).




For each comparison between spot beam and wide beam, the spot beam forward channel was less than or equal to the wide beam forward channel (both using 30 Msps).  Hence, the difference is negative or zero in all cases.   The variation from beam edge to beam peak is not significant in that the result is somewhat evident in all contours.

The following chart is the same conditions for the forward channel, but with reduced interference on the spot beam from the baseline. 





The differences are minimized, but still zero or negative.  The reduced interference increases the forward channel data rate for spot beams, but there is still more interference than with wide beams.

Spot beam data rates are lower than legacy wide beam data rates, but only if you assume a fixed bandwidth.

Spot beams support much higher symbol rates than legacy wide beam, which is the gateway to very big forward channels.

Spectral efficiency gain (spot - wide) reflects the same trend (scalable) to the change in data rate (since the analysis was done at 30 Msps), as shown in the following chart.








The following chart shows the same trends with reduced spot beam uplink interference.





As expected all forward channel spectral efficiency differences are zero or negative.

Spot beam forward channels are LESS efficient than wide beams.

Return Channel
The return channel bears a real benefit with the enhanced sensitivity of the satellite transponder. 

The following chart shows the difference in return data rate (spot - wide).






The differences are significant and largely positive.  The only time wide beam was faster than spot beam was the result of spot beam uplink interference.

Performance benefits from spot beams are impacted on the return channel by uplink interference.  

The following chart shows performance with reduced spot beam uplink interference.





The spot beam return channel has realized huge benefits or at least no worse with reduced uplink interference.

Spectral efficiency gains (spot - wide) follow the same (scalable) trends as data rate gains since using a constant 7.5 Msps carrier in all cases.







Spectral efficiency gains for reduced spot beam interference are shown below:





Return channel data rates are increased with spot beams.

Return channel spectral efficiency is increased with spot beams.


Performance

The following tables (baseline/reduced interference) document the estimated forward channel data rate comparing legacy to spot for each antenna and steering combination across five beam contours.

In each case, a carrier of 30 Msps was used operating at 10 dBW/4 kHz downlink PSD.








The following tables (baseline/reduced interference) document the estimated forward channel data rate using a spot beam for each antenna and steering combination across five beam contours.

In each case, a carrier of 230 Msps was used operating at 10 dBW/4 kHz downlink PSD.





It is quite conceivable for creating forward channel data rates, using a spot beam, well in excess of 100 Mbps.

The following tables (baseline/reduced interference) document the estimated return channel data rate comparing legacy to spot for each antenna and steering combination across five beam contours.

In each case, a carrier of 7.5 Msps was used operating at maximum PSD for the antenna and its steering command.





Only one return channel exists, creating a hard limit that can change with steering commands and airplane position and attitude.




Efficiency

The following charts show the range in forward and reverse (baseline/reduced interference) spectral efficiency changes as a function of beam contour and by antenna/steering combination.








Forward channel spectral efficiency ranges typically between about 0.4 to about 1.4 and are more a function of antenna than beam.










Return channel spectral efficiency ranges typically between about 0.1 to about 0.5 with legacy wide beams and about 0.8 to 1.8 with spot beams.





Capacity

Given a desire to deliver a 1 Mbps forward channel data rate coupled with a percentage on the return channel, how many 1 Mbps "personal services" can be supported simultaneously?

At $10 per hour retail pricing, a take rate of 5%, or perhaps less than 10 on a narrow body and less than 20 on a wide body.

In the following tables, three limits are noted.
  1. Return channel hard limit is encountered.  This is a greater issue with a low ratio forward to return (4:1 in particular).  These are shown with a white highlight.
  2. Wide beam forward channel data rate limit is encountered.  This is a hard limit assuming a single 36 MHz transponder.  These are shown with a red highlight.
  3. Spot beam forward channel data rate limit is encountered.  This is a soft limit, with the option to increase the channel bandwidth.  These are shown with a yellow highlight.
For each ratio (forward to return), two sets of tables (baseline reduced interference) are provided.  

The first set is based on a 30 Msps carrier.  

The second set is based on a 230 Msps carrier (spot beam only).  The 230 Msps carrier is large enough that the capacity is limited by the return channel. These capacity limits are a hard limit for spot beams.




4:1




At 4:1, the return channel limits capacity using wide beam  Spot beam capacity could support as many as 35-55 services.




8:1




At 8:1, the forward channel hits a hard limit on wide beam, wheres with additional bandwidth spot beam could support as many as 110 sessions. 





16:1







At 16:1, the forward channel hits a hard limit on wide beam, wheres with additional bandwidth spot beam could support as many as 221 sessions.

With a few exceptions, capacity is adequate to meet demand for $10/hour streaming sessions.

Capacity may be limited in disadvantaged steering solutions. 

Capacity is limited with legacy wide beams using 34x7 antenna.





Key Conclusions
  • Forward channel spot beam transponder available symbol rate is increased dramatically over traditional 36 MHz Ku-band transponders
    • Wide beam 36 MHz transponders are limited to about 70 Mbps
    • Spot beam 200 MHz transponders and over 300 Mbps are conceivable
    • Single forward channel may limit number of passenger sessions

  • The return channel benefits from spot beam transponder figure-of-merit (G/T)

  • Uplink carrier to interference is much higher with spot beams
    • Reduced spot beam forward channel spectral efficiency compared to wide beam
    • Moderates high transponder G/T return channel benefit from spot beams

  • Spot beam economic benefits are primarily a factor of spectrum pricing
    • Service favors forward channel 
      • where spot beam is slightly less efficient than wide beam
    • Assuming $3,500/MHz wide beam is based on industry norms
    • Assuming $2,000/MHz spot beam is on the premise that it should be cheaper
    • Cost savings is nearly the ratio of cost per MHz between spot beam and wide beam

  • For any optimized case (as assumed herein)
    • The forward channel always operates at higher spectral efficiency than the return channel
    • Forward channel should operates at downlink regulatory limitations
      • 10 dBW / 4 kHz in beam peak in all cases
    • There may be more than one forward channel
      • 30 Msps channel was assumed
      • 230 Msps full power channel (Spot beam only)
    • There is only one return channel
    • The return channel operates at uplink PSD regulatory limitations
      • simple antenna model used for analysis
    • The return channel may be limited by symbol rate
      • 7.5 Msps channel was assumed
    • The return channel may be limited by uplink EIRP
      • unlimited EIRP was permitted





Link Budget

A satellite beam has contours from the most powerful or most sensitive to the least powerful or least sensitive.

Forward Channel
The forward channel operates across some bandwidth and uses some amount of power, both of which are increased with spot beam satellites.  

For the purpose of this analysis, I selected 30 Msps (million symbols per second) which fits into a 36 MHz standard legacy transponder considering guard band using about 50 dBW.  There are 27, 54 MHz and 72 MHz legacy transponders.  

A spot beam transponder can offer higher bandwidths.  For example, a 58 dBW transponder could support a 230 Msps carrier.

Each beam may have access to two polarities and perhaps 500 MHz of capacity.  Spot beams offer larger bandwidth because of the ability to re-use frequencies like a cellular network.

In summary, a spot beam can drive higher bandwidth carriers.

Satellite regulations and coordination guidelines limit the power spectral density (PSD) in the downlink, which levies the same, hard limit (10 dBW/4 kHz) in any beam.  

The following table provides the downlink transponder settings for 30 Msps and 230 Msps and the downlink EIRP for the carrier across the beam contours.







Return Channel
The return channel depends on the satellite transponder to receive its uplink.  The figure of merit, or G/T, of the satellite transponder is significantly boosted with spot beams.

The following table provides the uplink transponder settings for both the legacy and spot beams.








Antenna under Test
Two different antennas were selected and modeled.

The 34"x7" multi-gimbaled array is meant to represent a typical Ku-band antenna in the marketplace today.  This antenna type is mechanically oriented such that the gain and beamwidths are constant. 

The 30"x30"  "flat panel" is meant to represent an emerging Ku-band antenna that sits flat to the fuselage.  This antenna type has diminishing gain and expanding elevation beamwidth as a function of elevation steering angle.  This antenna has higher gain than the 34x7 except at very low elevation angles.

Both antenna types suffer from skew angle effects.  The 34x7 skew effects are quite severe in comparison to the 30x30, even at low elevation angles. 

The following table provides key performance characteristics for the two antennas at two steering angles.






Two steering angles were selected:

60 deg elevation and 10 deg skew represents a favorable steering solution.  The 34x7 antenna has slightly higher discrimination on the GSO to adjacent satellites (adjacent channel interference) than the 30x30 antenna.  It should be noted that the 30x30 antenna performs markedly better at higher elevation angles in comparison.  

30 deg elevation and 30 deg skew represents a disadvantaged steering solution.  The 30x30 antenna has slightly higher discrimination on the GSO to adjacent satellites than the 34x7 antenna.  It should be noted that the 30x30 antenna performs markedly better at higher skew angles in comparison.

Using a common 7.5 Msps bandwidth for each return channel coupled with high PSD limits resulted in very high EIRP.  For this analysis, I assumed the power was available, but in fact some antennas would not be capable of such high EIRP.