Upcoming Maintenance Alert:

The UBNT Community will be upgraded at 5pm MDT on April 25th. During this time the community forums will be set to read-only status.

Learn more

×

Filtering in 3 Domains: Why PrismStation is the Greatest Product Ever Brought to the WISP Industry.

by Ubiquiti Employee ‎06-14-2017 09:03 AM - edited ‎06-14-2017 01:18 PM

 

1.jpg

 

Back in the early days, I was working with a customer to replace an older, slower 802.11b 2.4GHz mini-PCI radio with Ubiquiti’s latest SuperRange2 802.11g hi-power mini-PCI module.  Response to the SuperRange mini-PCI cards had been overwhelmingly positive, but this user was experiencing significantly worse performance post-upgrade.  At first I thought it was bad hardware and shipped another card, but the results did not change.  I then worked with him, trying to improve performance through several attempts at design modifications but also to no avail.

 

At that point, I bought a sample of the older 802.11b card at issue for myself, which was based on the Prism 2.5 chipset from Intersil.  And like the “Prism” name implies, I quickly saw that this older card although slower in max speed, had a radio design with a superior “selectivity” — the ability to filter out neighboring channels.

 

But, how could this be the case? The Super Range (Atheros 802.11a/b/g based) radios were the latest technology and I assumed it would outperform the older 802.11b technology in all areas. After taking apart the Intersil Prism radio, things became clearer.

 

The Intersil radio was based on a true “superheterodyne” receiver architecture where the carrier was down-converted to an Intermediate Frequency (IF) and filtered with a dedicated discrete filter. Meanwhile, the Atheros radio, was a completely integrated CMOS chipset without any off-chip IF filtering.

 

2.jpg

 

So what does this all mean? The ability to filter a radio signal largely depends on 2 things:

 

1. The fractional bandwidth:

 

Filtering out a 1MHz channel at 1GHz (.1% fractional bandwidth) is much harder than filtering out a 10MHz channel at 100MHz (10% fractional bandwidth)

 

3.jpg

 

2. The filter effectiveness:

 

A dedicated specialized filter is typically far superior than a filter integrated into an IC.

 

4.jpg

 

In the case of the Prism radio, it optimized both areas. By having a down-converted IF of 384MHz, it was able increase the filtering fractional bandwidth. And with a dedicated off-chip SAW (Surface Acoustic Wave) filter, it had a much more effective filter. In comparison, the Atheros radio was built for complete low-cost IC Integration and had neither. It might have performed well indoors, but in outdoor WISP applications, the Prism radio could survive in RF environments where the Atheros based radio had no chance.

 

Filtering in the Frequency Domain

 

This experience would plant the seed for what we ironically call our “Prism Technology” at Ubiquiti. We wanted a way to leverage the speeds of the latest WiFi chipset technology but also retain the great “selectivity” of the original Intersil Prism radios. 

 

What we patented would be counterintuitive to most. We essentially put our own radio in front of the WiFi chipset radio. How does this exactly improve performance? The diagram below helps explains the concept

 

 5.jpg

 

Our Prism technology receives the unlicensed band spectrum (5GHz or 2.4GHz), down converts to an intermediate frequency, applies specialized hi-selectivity filtering on the area of interest, and up converts the channel back to appear magically “clean” to the WiFi Radio.

 

We have proven selectivity improvements of up to 30dB. To put this in a linear perspective, our Prism Technology reduces noise seen by the Wifi radio by up to 1,000 times!

 

Filtering in the Spatial Domain

 

Horn antenna technology has been around for over a century, but only recently have they been attractive and proven to be successful in WISP applications. In the early days of this industry, antenna gain was most valued and traditional sectors and reflectors were best suited for deployments. Fast-forward to today, with billions of unlicensed radios in use worldwide, the ability of antenna isolation to mitigate noise is becoming more valuable. We want our antennas to only hear and talk in a single direction and “ignore” all other directions. Horn antennas do this exceptionally well.

 

The challenge for our antenna team at Ubiquiti was how do we take advantage of the RF isolation advantages of horns, but still maintain enough antenna gain required for high performance links?

 

The answer is what we call “asymmetrical horns” and we believe they are the future of WISP deployments

 

 6.jpg

 

Filtering in the Time Domain

 

The 802.11 WiFi standard uses something called CSMA/CA (Carrier Sense Multiple Access / Collision Avoidance). It is a contention based protocol which means if all clients on a network can “hear” each other, everything can work well. But, in the case of outdoor networks and isolated directional links, most of the clients cannot hear each other and end up talking over one another. To solve this we introduced a TDMA (Time Division Multiple Access) protocol where clients are assigned organized time windows to talk so they do not interfere with each other.

 

This was the essence of our AirMax TDMA protocol we have improved throughout the years. While this works well for clients connected to a single AP, what about interference issues with multiple APs co-located together? 

 

Our all-new GPS synchronization protocol specifically addresses this challenge. PrismStation uses GPS to provide a global synchronization timer for potentially every deployment in the world. What this means is multiple BaseStations can work seamlessly on a single tower or neighboring towers and even using the same spectrum. And we can also achieve synchronization between AirMax and AirFiber Basestations (including our upcoming LTU technology)

 7.jpg

 

We believe the culmination of technologies driving 3-Domain Filtering (Frequency, Spatial, Time) will enable the next stage of high-performance, high-density AirMax networks throughout the world.  We are really proud to bring this product to market and hope it will be the weapon that operators can use to fight and build higher performance networks even in the presence of increasing amounts of RF noise in the unlicensed bands.

Comments
by
on ‎06-14-2017 09:26 AM

We believe the culmination of technologies driving 3-Domain Filtering (Frequency, Spatial, Time)

 

Or, that line could read ... 3-Domain Filtering (Spectral, Spatial, Temporal) .... and then trademark Ubiquiti's SST (tm) filtering technology.

 

 

by
on ‎06-14-2017 09:56 AM

I have an unimportant question. Why was it named "PrismStation"? Station seems to indicate it is a client device like the nanostation, isostation, etc. instead of an AP or base... station... Ah, maybe that was why?

 

PrismAP seems like a better name of the radio, with some other name for the add-on horns.

 

by
on ‎06-14-2017 10:18 AM

@flipper- I think the marketing team just fell over themselves to get that to the relevant agencies.

 

Great read as these articles are.

by
on ‎06-14-2017 11:00 AM

The "-station" name comes from the IsoStation name - and NanoStation/PowerStation legacy names - Prism filtering (yes I like the SST thing too ) is mostly applicable for APs located at noisy sites and IsoStation Prism doesn't sound nearly so good... 

Jim

by
on ‎06-14-2017 11:02 AM

I love this.  It's really a genius idea filtering that way, and obviously works very, very well.

by
on ‎06-14-2017 11:03 AM

IsoPrism AP? Man Happy

 

by
‎06-14-2017 11:11 AM - edited ‎06-14-2017 11:11 AM

Really admire your writing skills, can explain technical stuff so well for everyone to understand. 

by
on ‎06-14-2017 12:43 PM

Looking forward to trying them out!

by
on ‎06-14-2017 12:49 PM

@flipper Love it..lol

Great product design!

by
on ‎06-14-2017 01:06 PM

So, what about 2.4GHz?

 

--

by
on ‎06-14-2017 01:13 PM
by
on ‎06-14-2017 01:31 PM

Simply put, a 2.4GHz horn would be approx. double the size of the 5GHz units, too big to be practical in most cases.   But you still get the other benefits of the Prism radios with the Rocket2AC Prism.

Jim

by
on ‎06-14-2017 01:39 PM
Sometimes the old tricks are the best especially with a new twist. Hearing that you are using superheterodyne techniques to acheive this awesome result just makes me feel warm all over. Great work Robert!
by
on ‎06-14-2017 01:39 PM

Ehi Bob, I'm with you!

by
on ‎06-14-2017 01:47 PM
I believe intersil patented the prism filter system for TV signals originally (wifi isnt the only crowded RF space). Im not sure who owns intersil nowadays.
by
on ‎06-14-2017 01:52 PM

Bought by Renesas in Febuary of this year, but still operating under their own name.   I looked seriously at using their Prism chipset and reference desogn to build our own 802.11b gear back in 1999.   Really glad I let folks like Robert do it instead ;-)

Jim

by
on ‎06-14-2017 02:09 PM

Horn's are nice, but not what I need...

 

I've been able to run with 120deg shielded sectors, 8pol filters, 10MHz channels and three AP's on a tower with minimal self interference.

 

To improve on my current system...

 

I need 2.4GHz with selectivity like the old Intersil gear...

I need working GPS...

I need backwards compatibility with my M clients, both XM and XW...

 

If I can sub these in for my existing Rocket 2M AP's and get those benefits, I am ready to buy!

 

--

by
on ‎06-14-2017 02:54 PM

Customers have quite a few expectations with this new generation v2, Congratulations Robert for the new product innovations

by
on ‎06-14-2017 04:50 PM

I tried 2.4 frequency. There were many problems. Did this product solve the jamming problem?

by
on ‎06-14-2017 04:54 PM

Define 'jamming problem' ...

by
on ‎06-14-2017 06:44 PM
Congratulations Robert, Tesla would be proud of you!
by
on ‎06-14-2017 10:18 PM

Robert,

 

What's the difference in size between a horn and parabolic antenna of equivalent gain (say 30db)?

 

Thanks,

 

Joe

by
on ‎06-14-2017 11:13 PM

Congratulations for discovering the SAW filters, other manufacturers have been using them for long time, but they don't call Airprism, just SAW filters.

 

Also if the GPS is finally working, 6 years late, let me know what to do with the Rocket M5+GPS I have installed in towers still waiting for a reliable firmware.

 

 

by
on ‎06-15-2017 03:04 AM

What is SST filtering?

by
on ‎06-15-2017 04:22 AM

The superheterodyne receiver was invented almost 100 years ago so hardly a new idea.
Are you using the filter on transmit (to reduce OOBE) as well?

by Ubiquiti Employee
on ‎06-15-2017 07:17 AM

@roanwifi

 

@david_lewis

 

Ha...I will expand a bit for you skeptics:

 

 

-Integrating a local oscillator that must "talk" to the WiFi radio to track carrier changes (both center frequency and channel width) and TX/RX changes. To do this over the complete 5GHz band with low phase noise and down conversion design that minimize intermod spurs is not a trivial task

 

-Yes, we use SAW filters...actually multiple SAW filters for different channel widths...5MHz, 10MHz, 20MHz, etc.  So our "selectivity" actually becomes even greater as you reduce channel width.  There are no SAW filters that operate effecitvley at 5GHz.  Downconverting to an IF enables the use of SAW filtering and we then upconvert  to the original carrier whic feeds into the WiFi Radio

 

-This is our 3rd generation of our Prism tech which has improved significantly over the years

 

 

-

 

by
‎06-15-2017 07:21 AM - edited ‎06-17-2017 11:26 AM

The PrismStation, Rocket Prism AC Gen2, and Rocket 5AC Prism have filtering on both TX and RX.

 

SST was @flipper's rework of Frequency, Space, and Time into Spectrum, Spatial, Temporal (first comment above).

by
on ‎06-15-2017 07:27 AM

this all sounds great but how many more years to we have to wait before these products are available at in the retail supply chain?

by
on ‎06-15-2017 07:34 AM

Not a sceptic, just amazed that direct conversion (cheap and you get what you paid for) is still used for “professional” equipment.

 

Are you using a VCO/PLL or DDS for the 5GHz local oscillator? Guess image rejection is probably an issue at 5GHz with an 384MHz IF?

 

Surprised there isn’t a chip set designed from the ground up as a superhet as down converting and then up converting will add noise.

by
on ‎06-15-2017 07:53 AM

As to availability, I'm installing them today...

Jim

by
on ‎06-15-2017 10:44 PM

As david_lewis said, downconverting, then upconverting to feed the wifi chips  degrades IP3

 

 

 

by
on ‎06-15-2017 11:31 PM

OFDM signal quality boils down to SNR which is degraded by;

  • Adjacent channel noise that is not filtered out
  • Noise in the frontend
  • Noise in the mixer
  • Noise from non-linearity (IP3)
  • Noise from the local oscillator(s)
  • Noise from the image

The issue to bear in mind is that a 6db improvement in SNR should double the range for the same speed.

by
‎07-09-2017 02:36 AM - edited ‎07-09-2017 02:37 AM

Black surface in front absosbs heat. I got one unit dead after 2 months uptime... We hit 45 Celsius under shade easy. I was forced to paint the rest units white lol. Tell your engineer to improve it !

by Ubiquiti Employee
on ‎07-09-2017 10:50 PM

Thanks for the feedback @tech53 - If you still have the dead unit, can you send it back to us so we can investigate and look into this further?    

by
on ‎07-10-2017 10:04 AM

sure.message me your address and I will ship it over from Europe

by
on ‎07-14-2017 10:54 AM

Incredible article.   I wish the rest of the world could be notified.   Sadly, I see it no where.   Find some amazing talented people that can tell the story of Ubiquiti.    If no one knows about it you are not maximizing your potential, or what you created does not exist. 

by
on ‎08-04-2017 05:28 AM

I want to use phones, tablets, laptops 2.4 in green color space.What can I do??

My map address... Mithamoin,Bazar, Kishorganj, BangladeshBangladesh_ptmp.png

by
on ‎08-04-2017 06:02 AM

@oajidul  You may want to look through the forums to find a more appropriate forum.  Maybe an airMax forum...

by
on ‎08-18-2017 07:16 AM

hi

 

My background is RF design and I have developed that exact thing before . We tended to refer to them as tracking filters.

 

 

The basics are two mixers and one Local oscillator plus Filter(s) .

 

You down convert with the first mixer, filter, then re-upconvert with the second mixer.  The local oscillator is the same for both so slight errors or drift (if it was a free running osc) was not an issue as long as it didn't take the I.F. outside the filters bandwidth.

 

I have worked on various "applications that cannot be named" that use this technique Man Happy

 

Bill

by
on ‎10-02-2017 08:20 PM

So am I reading this correctly, AirFiber and AirMax (Gen2) GPS Sync are compatible and can be colocated?

by
on ‎10-03-2017 05:57 AM

With the correct settings, they can share the same frequencies (with appropriate azimuth\elevation separation), but I wouldn't call them compatible. Maybe their sync methods can be compatible.

by
on ‎10-12-2017 09:38 PM

Anyone know the timeline of when these will be DFS compliant? I ordered 6 of them and was disappointed to find that DFS frequencies are unavailable. I absolutely love the design of these radios and I'm super excited to deploy them! Thanks Robert and team!