Sunday, February 28, 2016

Beam Hopping, Beam Forming, Frequency Reuse, and the Quest for Maximizing Satellite Throughput

Satellite communication is amazing in so many ways.

Start with launching a machine into space and expect it to operate precisely while managing to harvest solar energy and hold attitude and position.

Add to the mix operating at frequencies that frankly are absurd.

Wide-band transponders got us this far, but their utility now is just a gateway to the future of spot beams.

Spot beams are formed by higher gain antennas with smaller beam widths than continental wide beam transponders.  The beam width is a function of aperture and of frequency.  A given aperture delivers a smaller beam width with higher frequency.  A given frequency delivers a smaller beam width with a bigger aperture.

The last piece of the puzzle is the part that is in play, that of how to switch the information between the beams, and how many beams can you use?

Cellular technology embraces dividing the world into sectors, where each sector is assigned a pool of frequencies.  Interference is minimized by ensuring the adjacent sectors operate on different frequencies.  Re-use is accounted by how many sectors combine to use the assigned spectrum.  For example, in four-color reuse, the sum of the frequencies is divided into four pools, and these are assigned geometrically in a pattern where each border is across a different frequency pool.

Interference is a consequence of the reuse.  Dividing the sum of frequencies more and more leads to higher and higher immunity from interference.  Interference will influence the resultant demodulated signal which gets to spectral efficiency.

I find four-color reuse commonly used in broadband spot beam assignments.

In Ku-band, the sum of frequencies may be 500 -1000 MHz times two polarities divided by four equals 250-500 MHz per spot beam.  Global allocation is 500 Mhz. More than 1 GHz available in downlink to CONUS.

In Ka band, the sum of frequencies may be 500-1500 MHz time two polarities divided by four equals 250-750 MHz per spot beam.  Global allocation is 500 MHz.  ViaSat-1 uses as much as 1500 MHz downlink in CONUS, as an example.

Please note that the frequency assignment are a huge matter to satellite operators, that every licensee is looking for any and every slice that might have use, even if with difficulty.  For example, Ka band has a non-geostationary operator, O3B, which has 500 MHz that a geostationary operator, ViaSat, uses with the understanding that occasionally it will be unavailable as an O3B satellite transits through.  I don't profess to know what everyone can use.  It is clear 500 MHz is available in both Ku and Ka, and that both have access to other spectrum.

A fundamental challenge in four-color re-use is having the transmitting power to use the full spectrum available to a given spot beam, when considering that there are limits to how much power and hardware.  It is expensive, possibly wasteful, and ultimately limiting to build out every beam to a four-color reuse.  Driving innovation is the concept that demand is not homogenous, that demand shifts from beam to beam over the course of a day or seasonally.  Some beams may need the sum of the frequencies, not just a share. This has given rise to beam hopping.

As with every great technical challenge, multiple methods have emerged.  In this case, it is frequency-division Versus time-division.  Four-color reuse is a frequency-division method.  Beam-hopping is a time division method.

As with FDMA and TDMA (multiple access), in frequency-division the carrier is a fraction of the time-division carrier.  In other words, in frequency-division using 1/4 of the sum of frequencies all the time is equal in time-division using the sum of frequencies 1/4 of the time.

Beam-hopping has a companion technology which is the ability to switch the transmitting power from beam to beam as a function of time.  With beam hopping, each beam only needs the sum of frequencies for some portion of time, so each hopping transponder is plumbed into a succession of beams, dwelling just long enough to fill the demand in each beam.

Beam hopping may assign the sum of frequencies in one polarization to a set of transponders and the other polarization to another set of transponders.  For example, each transponder would have up to 500 MHz or maybe even 1500 MHz.  The number of transponders depends on how many spot beams and total capacity goals.  In general, there number of transponders totals to the data link capacity of the satellites.  The number of transponders is independent to the number of beams.  The satellite is final arbitrator, where the recent trend is trading electric propulsion for more payload and lifetime.  But there is only so much power regardless of what you want to do, and that limits the number of transponders (albeit, antenna gain is a factor).

In support of any broadband satellite venture are the service links to the remote terminals (to which this article is focused on), and the feeder links to the ground station that provides the connection to the Internet.  The challenge is that the feeder links do not steal from the service links.  Spot beams with frequency management gives rise to putting the feeder link into a different location than the corresponding service link.  This geographic diversity, if managed, can offer a complete service utility of the forward and return channels.  Alternatively, the feeder links can operate on even more absurd frequency bands like Q/V where about four GHz is available in both uplink and downlink.  Then there is optical.  The bottom line, a given ground station can deliver at least 6 Gbps to the forward service link.  If you want to aggregate hundreds of Gbps you will need to aggregate many geographically diverse ground stations.  This ground station complexity can circle into the modem, making the modem highly customized to a given network.  Operating with two broad band satellite networks may give rise to needing to support two modems.

In frequency division, the spot beams are designed to have a defined border area such that a degree of isolation exists between each sector.  With beam hopping, many more spot beams can be supported with overlapping patterns.  The benefit to more spot beams is putting the beam center onto each remote terminal to drive the strongest signal.  In frequency division, the remote operates across a few dB of variation from beam center to beam edge.  In beam-hopping, the remote may always operate near beam center.

In frequency division, the total capacity to a given spot beam is limited to a fraction of the sum of available frequencies.  Furthermore, the capacity may not be distributed to meet unforeseen demand. With beam hopping, the sum of frequencies can be brought to any beam, and furthermore each remote may operate at a higher efficiency level.  If not for the interference.

We casually assume spectral efficient of 2:1 when converting from bps to Hz; e.g 1000 Mbps takes 500 MHz.

Aeronautical terminals are inherently disadvantaged.  2:1 is a difficult target to achieve.  I will leave this aspect to the side from this discussion.

Here are some rules of thumb (forward channel)

Seven color reuse:  C/I 18 dB, resultant C/N 12 dB, SE 2.8
Four color reuse: C/I 10 dB, resultant C/N 6 dB, SE 2.0
Two color reuse; C/I 1 dB, resultant C/N 2 dB, SE 1.2

In four color reuse, one beam of 750 MHz delivers about 1.5 Gbps

In two color reuse, one beam of 1.5 GHz delivers about 1.8 Gbps

In beam hopping, it is possible for both polarities to dwell on the same beam at the same time.  In this context, one beam could receive 3.6 Gbps (optimistically).

Frequency-division is a relatively simple concept to grasp and model, but inherently limited.

Beam hopping can deliver the service in a more flexible manner.  The spot beam architecture of a matrix of overlapping patterns allows in-orbit response to demand as it shifts, or if the satellite is repurposed.  Phased array technology permits more and more flexible beam-forming, which translates into even greater in-orbit flexibility.

While it may be easy to utilize apertures around 2 meters, the use of unfurling apertures spanning 5 meters to even more then 9 meters are well underway.  A 5 meter aperture will more than triple the number of beams from a 2.6 meter aperture.

Harris Ka-band Unfurling Antennas

In summary, expect a movement in both Ku band and Ka band towards smaller and smaller spot beams and using beam hopping to shift the service to match the demand.  

The remote terminals benefit from the best antenna technology coupled with modems that are matched to the needs of beam hopping, notably very high symbol rates (300 MSps or greater) to support the wide carriers. It may be necessary to carry more than one modem.

The satellites become more and more standardized, with in-orbit beam forming and beam hopping.  

It is my observation that Ku-band service providers are professing a movement towards a commercial offering using a generic satellite design with in-orbit inter-operable configuration, whereas Ka-band service providers are building more customized satellite networks around complete vertical integration.  

Peter Lemme

Copyright 2016
All Rights Reserved.