Communication Network model

Who's going to fabricate the hardware for your new network model?

I'm with CptOmily in that unless your a govt entity or a company like IBM or Microsoft where in a newly emergin industry like computers you developed the standards for equipment and software that evey company and individual will want, which over time becomes the standard for all, creating your own communications model for computers is pretty moot. However, to give you resources in which you can research yourself, The International Organization of Standards [ISO] typically sets the design standards to which all others "should" conform and those design standards come from the companies [or governments] who initially created it not because they had the foresight to see what that industry could grow into but because of an immediate need.

Communication Models ?

Why not adapt the microsoft network model / linux models as the starting point ? You are right, OSI and TCP/IP are network models. OSI is more general. It may not be what you need to consider.

Defines what may be needed first to build a communication network:

1. Communication Medium - Physical Layer - Fiber Optic, Wireless, Coaxial Cable or CAT5/CAT5e/CAT6 cable. Check out the physical layer protocols.

2. Hub - Switch - Router: Do you need a signal Repeater or Switch (Store and Forward) or Router (w/ routering tables). What about VLAN ?

3. Do you run pure DOS program on your network ?

4. Do you need Internet Connection ?

thanks allot'theohkm@hotmail.com'..This kind of answers I need.. I'm making a research how IT people from different levels will answer.. but still I need more... Look at the layers of two references TCP/IP and OSI.. and give theoretical answer...how you can develop your own reference model...I'm not planning to implement the network...

If I understand what you're trying to do correctly, I would go back to first principles and think about why people want a model like OSI in the first place. My main interpretations of this question are:

The different levels perform different functions in the processes required for communication.
Having separate elements for each layers - but with a standard interface between the layers - means that if your requirements change at each layer, you don't need to change all the layers. (e.g. IP will happily run over Frame-Relay, ATM, PPP etc.)
Having different layers means that you can reduce processing at intermediate network points - they shouldn't need to process all the layers. (routers usually only process to Layer-3)
The layered approach makes things easier to understand.
I'm sure there are other advantages I've forgotten, but you see my point

You therefore should start by defining all the things you want your model to achieve and break them up into a logical order in which these functions should be implemented, which you can group together and which should be separate etc.

Other people mentioned the other models and their starting points. The problem you may find is how to avoid just copying what they've done (there's a good reason why everyone uses them - they work!)

Here's an alternative challenge for you: modern networking uses so many protocols and runs multiple protocols over other protocols over other protocols, that large amounts of the data flying around the world consists of overhead. How about a model that aims to reduce the required overhead?

Best of luck,
Paul

CCDA, WLAN spec.

I think people are confusing the issue, primarily by focusing on current standards instead of the parameters that drove the standards.

Key parameters that any communications network must deal with include:
Function - peer to peer, one to many, or many to many? Converse or merely deliver? 100% accuracy or less?
Scale - important if addressing is a need.
Direct or indirect? Impacts routing.
Fixed points or mobile points?
Static or dynamic operation?
Traffic pattern (volumes, directions).
Control (fixed or user selectable reception).
Security.
Reliability.
Technology availability and constraint (cabling, RF, optic, interference concerns).
Interoptibility.
Cost and value.

Some examples of non OSI or TCP/IP networks to review include TV broadcasts (one to many), one way paging (one to any), telephone (peer to peer, modified to one to many).

Variations include FDM where the frequency determines who receives (from HF through optic DWM) where the tuned frequency can be the address, telephone where spanning tree and now number databases determine end points.