Patent question

[activeco]

Hello all,

This is a kind of non-technical, but still related question.

Does anyone have experience or some knowledge in patenting software solutions regarding ethernet?
Particulary I would like to know if it would be possible to claim a patent related to (nearly complete) elimination of ethernet collisions, based on new headers' definition.
The "problem" could be that one of the protocols, the tcp/ip is an "open source" issue and as such impossible to protect, even in commercial aplications.

Any comment is welcome.

 

[jimbopalmer]

Ethernet collisions were eleminated a decade ago by full duplex switching.



I tried to remain child-like, all I acheived was childish.

 

[activeco]

Thanks for your reply jimbopalmer.

I am afraid you didn't read the question carefully.
I was talking about software/algo solution, thus eliminating hardware.
Besides, even by using switches, you still have (smaller) collision domains, unless you refer to the home networks where every node could afford dedicated switch port.

 

[pansophic]

If you are claiming that you can eliminate collisions based on modifying the Ethernet header, then the fact that TCP/IP may ride on ethernet actually has nothing to do with your patent claim.

If your claim is that you can eliminate collisions based on modifying the Ethernet and IP, or Ethernet and IP and TCP headers, then you are better off issuing an RFC to the IETF and claiming your fame that way.

Since this problem has been solved for many years in hardware, it is unlikely that the patent will have any value, it is simply for bragging rights.

BTW, I haven't been in a business environment in many years where every port wasn't switched. Small, medium and very large. As a matter of fact, we have to search for non-switched devices to run simple packet collection applications. Most of those devices are 10 Mb/s hubs.


pansophic

 

[activeco]

If you are claiming that you can eliminate collisions based on modifying the Ethernet header, then the fact that TCP/IP may ride on ethernet actually has nothing to do with your patent claim.

Probably, but many lawyers would strongly disagree with pure logic.

Since this problem has been solved for many years in hardware, it is unlikely that the patent will have any value, it is simply for bragging rights.

This is also a very interesting point.
Is it a general opinion that such hardware savings don't deserve a kind of "radical" protocol changes?

 

[franklin97355]

You need to contact a patent attorney with these questions.

 

[pansophic]

Probably, but many lawyers would strongly disagree with pure logic.

Then you simply apply the patented technology to IPX/SPX or Appletalk instead of TCP/IP. The collision avoidance will be the same, and there is no question that the protocols are wholly independent. Ethernet is simply a layer 2 transport protocol, independent of the lower and higher level protocols. Since that can be simply demonstrated and is well documented, any lawyer worth 2 cents should be able to successfully argue it.

Is it a general opinion that such hardware savings don't deserve a kind of "radical" protocol changes?

It is more a matter that the second solution to the same problem is rarely effective in the marketplace, even if it is a superior solution.

But protocol modifications are extremely hard to implement because all devices in the network have to be modified in order to even test it. Switching was effective because it could be implemented in a single device, without any modification to its peers or dependent devices. You could replace one hub in your network with a switch, and no other modifications were required. People will nearly always choose this type of option over a wholesale upgrade. IPv6 is a prime example. There would be no issues on the current Internet with network numbering using IPv6, but it requires that all devices on the network run it, and it requires that gateways be installed that can covert back to IPv4. Therefore its implementation is extremely limited.

But it doesn't mean that the next generation of Ethernet devices couldn't implement your collision avoidance if there was a way to implement it on capable devices and not on incapable devices. This was accomplished when 100 Mb Ethernet was introduced, and then full-duplex Etherenet. A simple negotiation was performed on activation of the devices, with a logical fallback mechanism.


pansophic

 

[jimbopalmer]

Yes, a software solution would have been great 15 years ago, before switches were everywhere.

I have not dealt with ANY business in the last 5 years that was not deploying switches everywhere.


I tried to remain child-like, all I acheived was childish.

 

[activeco]

OK guys, thanks verymuch for your inputs.

 

[activeco]

OK guys, thanks very much for your inputs.