Josh, you say “the formulas are at the top”, and I respond that my complaints are about the USE of the formulas and cite 3 specific complaints about the numbers plugged into the formulas and also my complaint about the result saying 802.11n is only capable of 24 Mbps in a 20 MHz channel when we know the number is higher (around 75).
You respond “then post the correct formula”. The only conclusion I can draw is you didn’t read my post. Which is fine, but why respond to something you didn’t read? The thing about formulas, you can get all sorts of answers plugging in different numbers. The error is usually in the assumptions, not the formulas. For example, when someone announces a breakthrough in throughput and range, often it’s because they assumed extremely high S/N ratios, in other words no interference or thermal noise. That’s not a breakthrough, it’s a result of assuming away the fundamental challenge of RF communications. C=B*log2(1+S/N) says if you assume infinite signal or negligible noise, infinite channel capacity is possible. But that assumption is not relevant to the real world, where interference exists, thermal noise floor exists, and there are regulatory limits on transmit power. This is just the most common example of the fault being in the assumptions, not the formula. I think the most questionable assumption in the case at hand is applying a frequency reuse factor of 1/3 to 802.11n and 1 to LTE-A and comparing those as if they were both cellular systems. In the WISP world, when we cite spectral efficiency, I don’t think it’s standard to apply a frequency reuse factor. We just divide bitrate by channel width. Also questionable is comparing 802.11n vs LTE-A as if that was an apples-to-apples comparison. At a minimum, the comparison should be to 802.11ac, which is left out of the chart. Furthermore, there are assumptions behind the performance of LTE-A that don’t apply to wireless LAN, the most obvious being LTE-A is used in licensed, exclusive use spectrum, where interference comes only from other sectors or cells operated by the same provider. Similarly, if you look at Part 101 licensed backhauls you will see modulation levels like 2048 and 4096 QAM. Why does 802.11 not use such high modulations? Because the S/N to achieve those modulations is unrealistic in the environment where 802.11 systems are used. Also, if you just look at standards, 802.11ac is capable of up to 8 spatial streams also. The question is how to utilize that in a WISP environment. I think we see Mimosa trying, or at least they are trying to use 4 streams. But you also see Mimosa taking a stand against the FCC out of band emissions change. (did you read their filings?) I think it’s dangerous to say go ahead and require 20+ dB more OOB filtering, we just need to think out of the box and come up with new modulation schemes to make up for the lost throughput. Heck, we’re going to need all those advances just for more throughput, not to compensate for a filtering requirement that is totally unnecessary. From: Josh Reynolds via Af Sent: Monday, October 27, 2014 12:02 AM To: [email protected] Subject: Re: [AFMUG] questions about filters Wikipedia, and other resources, are what people let them be. Ken made some valid points, and I even said that -- twice. People who have been in IT/telecom for a long time get a certain attitude about them, and normally it's not a helpful one to people who might be able to learn something from "the old farts", but only if said individuals are willing to take the time to educate. Like I said, I don't think there's anything wrong with saying "then post the correct formula, in your humble opinion, so it can be fixed". Josh Reynolds, Chief Information Officer SPITwSPOTS, www.spitwspots.com On 10/26/2014 08:46 PM, George Skorup (Cyber Broadcasting) via Af wrote: Josh, you have strong opinions and there's nothing wrong with that, but at times you come off very confrontational, IMO. Ken is one of the smartest people I know and I have great respect for him. I think most others here would agree. On 10/26/2014 11:28 PM, Josh Reynolds via Af wrote: If you're not fixing to the problem, you're contributing to it. You have some valid points about weaknesses in the formulas used in that chart. Do you talk to everyone this way? Josh Reynolds, Chief Information Officer SPITwSPOTS, www.spitwspots.com On 10/26/2014 07:16 PM, Ken Hohhof via Af wrote: Are you trying to be annoying, or just succeeding? From: Josh Reynolds via Af Sent: Sunday, October 26, 2014 10:14 PM To: [email protected] Subject: Re: [AFMUG] questions about filters Then post the correct formula, IYHO, so it can be fixed. Josh Reynolds, Chief Information Officer SPITwSPOTS, www.spitwspots.com On 10/26/2014 06:20 PM, Ken Hohhof via Af wrote: That doesn’t address my complaints about the USE of those formulas. Do you agree that WiFi bits/sec/Hz should be divided by 3 but LTE should not, because of assumptions about frequency reuse? In the context of a WISP application which may use GPS sync? How about assuming one spatial stream for WiFi but 8 for LTE? And what about treating LTE Advanced like a current technology but 802.11ac as a future technology? And 802.11n is capable of more than 1.2 bits/sec/Hz. If the formula disagrees with reality, it’s the formula (or the numbers plugged into the formula) that must change, not reality. It’s not like a Looney Toons cartoon where the character falls to the ground once you point out they can’t walk on air. From: Josh Reynolds via Af Sent: Sunday, October 26, 2014 8:59 PM To: [email protected] Subject: Re: [AFMUG] questions about filters The formulas are at the top of the chart. Josh Reynolds, Chief Information Officer SPITwSPOTS, www.spitwspots.com On 10/26/2014 05:31 PM, Ken Hohhof via Af wrote: I think those numbers are flawed. Especially dividing the 802.11n numbers by 3 due to “frequency reuse” factor. And using SISO for 802.11n but 8x8 MIMO for LTE. Not to mention using 802.11n and not 802.11ac. Saying 802.11n is only good for 1.2 bits/sec/Hz is saying it can only do 24 Mbps in a 20 MHz channel. Hogwash. From: Josh Reynolds via Af Sent: Sunday, October 26, 2014 5:49 PM To: [email protected] Subject: Re: [AFMUG] questions about filters Well... http://en.wikipedia.org/wiki/Spectral_efficiency 802.11n has a spectral efficiency of around 1.2. LTE advanced has a spectral efficiency of _30_. If we could get some fairly cheap radio chipsets with even a 10-15 in spectral efficiency at this point, we would probably all be incredibly happy. Doing that would likely cause us to (A) Not be compatible with 802.11 (fine by me), and (B) would require mass market adoption. Josh Reynolds, Chief Information Officer SPITwSPOTS, www.spitwspots.com On 10/26/2014 02:40 PM, Mike Hammett via Af wrote: That's what I was hoping for but I was told to sit down. ----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com ----- Original Message ----- From: "Bill Prince via Af" mailto:[email protected] To: [email protected] Sent: Sunday, October 26, 2014 12:36:58 PM Subject: Re: [AFMUG] questions about filters Perhaps some innovation in improving efficiency? Maybe takes someone thinking outside of the current box(es). bp On 10/26/2014 9:55 AM, Chuck McCown via Af wrote: I was just going to mention that. Make a clean signal and you don’t have to filter so much. Anyone remember what a Class A amplifier is? (45% efficient at best) Cavity filters? I would think that in this day and age, you ought to be able to go DSP direct to antenna up to a 5 volt p-p signal. Or if you had to use a PA, inject a pre-distortion component. The cable TV guys have been dealing with these issues for decades. And then there is the issue with physical size of filters. A nice filter, with decent response and low insertion loss is large. SAW filters are about as small as you can get but they are higher loss than, for example, a waveguide filter however they are maybe 1% of the volume. You want a small radio that consumes very little power, then ... it will be more noisy than a large radio that consumes more power. That said, modern tech is unbelievable in performance and it just keeps getting better. Perhaps Chuck will get to come to AnimalFarm this year and show us something fun. From: Chuck Macenski via Af Sent: Sunday, October 26, 2014 10:24 AM To: [email protected] Subject: Re: [AFMUG] questions about filters Hi again, Another factor that causes expense is the linearity of the final stage output amplifiers...these puppies are linear for most modern radios and more linearity = more cost and higher power consumption. I will stop now... Chuck On Sun, Oct 26, 2014 at 10:47 AM, Chuck Macenski < [email protected] > wrote: Hi, There are many questions (explicit and implicit) in your question. Focusing on the tx side only (since we are talking about band edge), the filters you are talking about are electromechanical. Do a wikipedia search on SAW filters and you will get a sense for what you are dealing with. There are many other factors involved in meeting band edge requirements and other filtering that is or can be performed, but, the expense is often in the electromechanical components. Chuck On Sun, Oct 26, 2014 at 1:45 AM, That One Guy via Af < [email protected] > wrote: with the changes in the 5ghz rules, it may force innovation in filtering technology to bring cost down, assuming the innovators arent stuck in a mindset of the only thing that would work is what there is. How do filters works? Are there electronically adjustable filters? Where does the cost come from on filters? It is not new technology, so recovery of R&D on a new tech has long since past, what is it that drives the cost up? Is it primarily a matter of it being something needed, so its more valuable, or is it something in the physical properties of the filters that drives up the cost? Can you filter electronically a transmitter using something along the same lines of noise cancelling headphones
