<font face="arial" size="3"><p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;">Interesting ideas, but these ideas don't belong in TCP.</p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;"> </p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;">It's perfectly OK to have acknowledgements at the MAC layer. After all, it affects a single contention domain, which is like a shared link, not an end-to-end connection.</p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;"> </p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;">TCP's flow and congestion control mechanisms are conceived entirely around the idea of a heterogeneous dynamically changing overlay graph.  The expectations of the layer below IP are simple: no buildup of latency, reasonably high reliability (this is what "best efforts" has always meant).</p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;"> </p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;">The idea of making TCP the "physical layer" protocol was NOT in the original design, on purpose! I was part of the design team working directly under Vint Cerf and Jon Postel when this was decided.</p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;"> </p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;">Instead, the idea was that the physical layer properties (including "medium acquisition time", better called "arbitration overhead") would NOT be visible at the end-to-end level.</p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;"> </p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;">So, I would just suggest that these experiments demonstrate how to think about making *WiFi* fast, and not how to make TCP special over WiFi.</p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;"> </p>
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;">On Friday, May 31, 2019 11:38am, "Dave Taht" <dave.taht@gmail.com> said:<br /><br /></p>
<div id="SafeStyles1559322403">
<p style="margin:0;padding:0;font-family: arial; font-size: 12pt; overflow-wrap: break-word;">> http://www0.cs.ucl.ac.uk/staff/m.handley/papers/hack2014.pdf<br />> <br />> --<br />> <br />> Dave Täht<br />> CTO, TekLibre, LLC<br />> http://www.teklibre.com<br />> Tel: 1-831-205-9740<br />> _______________________________________________<br />> Make-wifi-fast mailing list<br />> Make-wifi-fast@lists.bufferbloat.net<br />> https://lists.bufferbloat.net/listinfo/make-wifi-fast</p>
</div></font>