Re: [lwip-users] Wireless TCPIP

[Ashutosh Srivastava]

Thanks for the response Gustl. I raised this issue, because yesterday I was
brushing up my basics from Computer Networks by Tannenbaum, and one
section described that TCPIP stack may involve some changes for wireless
interface, particularly congestion control.

It says, that TCP implementations assumes that timeouts are caused by
congestion, not by lost packets. Consequently, when timer goes off, TCP
slows down and sends data less vigorously to reduce congestion.

But it case of wireless, there could be packet losses and proper approach
in dealing with lost packets is to send them as quickly as possible. Slowing
down will make the matter worse.

Is the above statement true?

Thanks,
Ashutosh

----- Original Message ----- 
From: "Bernhard 'Gustl' Bauer" <address@hidden>
To: "Mailing list for lwIP users" <address@hidden>
Sent: Friday, November 11, 2005 2:02 PM
Subject: Re: [lwip-users] Wireless TCPIP


> Ashutosh Srivastava wrote:
>
> > I have successfully interfaced LWIP to my system with an ethernet network 
> > interface. Next on line is porting this to a wireless network interface.
> > My question is, apart from the underlying driver changes, are there any 
> > differences in TCPIP stack itself for it work with wireless interface?
>
> Hi,
>
> I've never programmed anyting connected with wireless. But I remember an 
> issu that came up with wireless and bridge code. IIRC the problem was that 
> wireless IF are not capable of receiving packets with an unknown MAC, but 
> bridge code changes the sender MAC. I solved this problem by setting up a 
> tunnel between 2 wireless IF and the bridge code used the 2 tunnel 
> endpoints.
>
> What I want to say is that I think there is now difference in tcpip, but 
> in the arp handling.
>
> HTH
>
> Gustl
>
>
>
> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/lwip-users