From: by way of Rohit Khare (Reiner.Ludwig@ericsson.com)
Date: Mon Jul 17 2000 - 11:13:15 PDT
my dissertation titled "Eliminating Inefficient Cross-Layer
Interactions in Wireless Networking" is now available at
http://iceberg.cs.berkeley.edu/papers/Ludwig-Diss00/index.html
The main contributions related to end2end:
** Investigation of interactions between link layer and TCP's error recovery:
(A) We show that highly persistent link layer error recovery is
required to optimize the end-to-end performance provided by
fully-reliable flows (e.g., based on TCP) while efficiently utilizing
radio resources.
(B) We show that TCP needs to be made more robust against spurious
timeouts and packet re-orderings. As a solution we propose an
enhancement to TCP's error recovery scheme, which we call the "Eifel
algorithm". The Eifel algorithm also provides for the implementation
of optimistic retransmission timers, because it reduces the penalty
of underestimating the round-trip time to a single spurious
retransmission (in the common case).
** TCP's RTT estimation:
(A) We reveal four major problems with the current de facto
implementation of TCP's retransmission timer.
(B) We propose a new retransmission timer, which we call the "Eifel
retransmission timer", that eliminates those problems. In addition,
the new timer takes advantage of the Eifel algorithm by becoming
increasingly optimistic while adapting to the measured fraction of
spurious timeouts.
** Link layer design philosophies:
(A) We show that the commonly used link layer design philosophy
"leave the link layer dumb but simple", causes reduced end-to-end
performance due to inefficient cross-layer interactions and a waste
of radio resources. Note, that the "end-to-end argument" [SRC84] is
usually quoted to advocate this design philosophy.
(B) The concept of flow-adaptive wireless links is a new link layer
design philosophy we have proposed. It eliminates all known
inefficient cross-layer interactions, with the exception of the
problem of competing error recovery which we solved with the Eifel
algorithm. We argued why carrying a network end-point's QoS
requirements as part of the flow's packet headers and accordingly
adapting link layer error control, is orthogonal to the "end-to-end
argument". Moreover, our solution has the key advantage that it
avoids performance enhancing proxies and their drawbacks.
///Reiner
This archive was generated by hypermail 2b29 : Tue Jul 18 2000 - 03:22:17 PDT