[Wormnet] Please start to play please, chips are in market to do!

[Peter Steen]

Hi Folks!

Bandpass filters, huh thats low :-)

Presentation of myself
----------------------
Well I do not consider myself to be qualified as joung scientist, not even 
scientist but well engineer with focus on Unix, TCP/IP and Datacoms + Telecoms.
I have over 250,000 hours experience of Unix servers since 1988.

My speciality is the space engineering, the space segment. I the real life I 
work as contractor selling competense to various companies (Business 2 
Business). About 40 of these contracts each year.

We educate people in unix basics, server, networking, enterprice enviroments. 
Mainly Solaris but also other unixes like linux and bsd family.

Also we sell various electronics kits like FM transmitters, remote sensors, lie 
detectors TV links and other nice kits.

We run two large wireless networks for internet connection with many customers 
inside since year 2002. They are baes on WiFi technology. Network is called 
blue bucket because the first access points were built inside blue buckets.

Windows, well good as workstation and playing games with. It is a very 
expencive solution in comparasion with unix solutions. We calculated about 
12,000SEK/workstations per year in our organisation between 1994 to 2002. All 
unexpected stops and updates included.

Unix, well we have one heavely used server up since 2001 with workstations, it 
was never unexpected stopped or rebooted. It just works! 2200SEK/workstations 
per year it have so far.

Fire pump
---------
I think we sit in the situation like when the liquid rocket was in development, 
the turbo pumps were to be designed and nobody did how. A meeting was set up 
with pump experts and interested, one person from the audience raises and spoke 
out, what you describe is a Fire Pump for water. 

Hope some here in this meeting can help telling what kind of "filter" we 
need if filter at all.

I think the filter is the second step, and we should first try to see if the 
protocols works, there are many questions.

Filters discussion
------------------
But anyway About filters for m8114)))))

First it is nessesary to know types exist and for what they are good.

We have on the shelf:

First in theory there are two kinds)
 Passive and Active filters.

Passive filters are made up from basic electronic and mechanic components.
Active filters can be as simple as you just att some dB to the signal in order 
to compensate for filter losses or more complicated feedback circuits.

When it comes to BANDPASS filters.
 Resonant bandpass or band stop(notch) filters.
 Bandpass filters made up from two others, lowpass plus high pass filter.

Dependent on what you are supposed to do.
 RC filters
  Made up from resitors and capacistors. (good for low to medium frequencies.)
 LC filters (good for medium to high frequencies)
  Made up from inductors and capacistors.
 Microwave filters can be made from copper/gold stubs and small resonant boxes 
made up from copper or gold.
 Software filters. (dependent on your math processing capacity)
  All the above can be "simulated" with discrete mathematics like z-transforms 
and similar.
  FFT fast fouirer transforms, fouier analysis of a spectra.  (dependent on 
your math processing capacity)

Next is to know what frequency you need to pass. If it should be tuned a lot
of problems apply, especially if it is over a large spectra and if it shuld be 
fast.

To make it more complicated before bringing in some maths at all we need to 
know the filter passband charateristics, namely the slopes derivata usally 
written in dB per octave.

All filters are more built up from "modules" like one resonant circuit with a 
coil of certain Q, the Q makes the filter more or less "flat". These can reach 
really steep slopes but to a price, they start ring!

It can also be built up from plain LP/HP LC stage or RC stages, these are 
cascade coupled to reach steep slopes. Usally they are 3dB/octave as best for 
one stage.

Now we are back to the narrow band, how narrow, 1Hz, 1KHz, 10KHz or any other?
The more narrow the more steep slopes need to be = more components or more 
processing power. Practical values can be 20-60dB/octave = many cascade stages 
or very good Q value or a very exact microwave cavity.

Then add to it to be able tuning the cascade, it is practical for small bands 
like the classic LW, MW, UKW, VHF, UHF and microwave bands, only because it 
is "narrow" maby some 1-2GHz as maxim. To Sweep the whole band very fast will 
require parallell stages if using the classic designs. Also the lower 
frequencies need large metric sizes!

I will say for wormnet all the classic filter designs are disqualified because 
they are not easely tuned over so large frequency spectra all the way down from 
some hundred kilohertz up to some gigahertz. Maby it can be used for some entry 
parts if system need to split up spectra in some bands of practical reasons.

It would be very nice to process whole incoming spectra directly with heavely 
modified FFT, in same time you would need more processing power than yet has 
been achieved.

All designs of today split up the spectra in narrow bands, with frontend 
classic filter designs. Then they mix down the selected spectra with non linear 
component like a diode or class b amplifier to a more comfortable band at lower 
frequency. It is then possible to process the spectra with FFT or other CPU 
based mechanismes. It is not a fast process to do so and also it will be a 
heavy construction with many components, also it will eat power.

It is from the beginning nessesary now to decide in what spectra wormnet should 
work. First after that it is possible to start thinking about the ENTRY SYSTEM 
for wormnet. Because it was never specified before. You cant say whole spectra, 
because from 0Hz to eternity does not apply.

Also Wormnet is a bit to classic in its ideas, it is made up from sinusidal 
waves all in whole classic radio communication, there is not into discussion 
wavelets or dirarc pikes which also can beused to communicate with. Then the 
filter discussion is also a bit "AGED". Wormnet was not specified for which 
communication methodes it should use in detail.

The dialogue of "noice summa-difference" communication is a bit aged, it is not 
function for higher frequencies or at longer distances. There are many more 
things to drop in wormnet. The first one is classic radio thinking!

I can put out the formulas for passive and active filters and software examples 
for FFT I have here as links to my-site and my-othersite. I did never put those 
out because I did not ever imagine one could be interested.

Cips in market
--------------
My recommendation is that we start this project using some of the existing ISM 
bands and by using existing components that exists from Maxim and nRF and Texas 
Instruments and Analog Devices and Chipcon and Motorola. Also I recomend using 
any GPS cheap modules that exist out there. I use Mitsubishi module, size is 
like 1cm times 1cm.

I have in person tested and sucessfully built one Videolink and one FM stereo 
transmitter and one reciever with components from Maxim. Also I have deeply 
investigated Chipcon products which are very sensitice in some cases close to 
space probe designs. Many of those includes the lawful transiever and also a 
microcontroller.

Wormnet protocols
-----------------
Using existing chipsets we can start develope the fundamental wormnet 
communication protocol:
Neighbour discover
Neighbour position: LIDAR GSM
Neighbour contact
Neighbour linking handshake
Neighbour link quality
Bouncing communication
Roaming protocols
Paralell transfers
Swarming data
Peering data
Anycast, Unicast, Multicast


Regards //
       //  Peter Steen


-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/