Changeset 2788

Show
Ignore:
Timestamp:
10/30/09 16:08:05 (4 weeks ago)
Author:
stevedh
Message:
  • minor nits, arsalan's changes
Location:
docs/Lowthane/ipsn10
Files:
3 modified

Legend:

Unmodified
Added
Removed

  • docs/Lowthane/ipsn10/evaluation.tex

    r2785 r2788  
    55performance in the literature.  Since we embrace the IP architecture proposed 
    66by Hui and extend his routing mechanisms, this work is the first in-depth 
    7 analysis the performance of a protocol which embraces a layered IPv6 
     7analysis of the performance of a protocol which embraces a layered IPv6 
    88architecture in this class of networks. 
    99 
    … …  
    2323\begin{description}\addtolength{\itemsep}{-0.5\baselineskip} 
    2424 
    25 \item[Reliability]  Does the protocol successfully deliver packets? 
     25\item[Reliability]  How well does the protocol successfully deliver packets? 
    2626 
    2727% The lossy nature of wireless mesh networks means 
    … …  
    5757%evaluate the stretch incurred using {\lowthane}. 
    5858 
    59 \item[Agility/Stability]  How quickly can {\lowthane} respond to changes in 
     59\item[Agility/Stability]  How quickly can {\lowthane} respond to failures and changes in 
    6060topology? 
    6161 
    … …  
    117117Testbed B & Motelab, Harvard \cite{WASW05} & 128 & 8 & Two hours & Office environment\\ 
    118118Deployment 1 & ACme \cite{acme} & 57 & 8 & Six months & Office environment \\ 
    119 Deployment 2 & Hydrowatch 2 \cite{hydrowatch} & 32 & 5 & Two weeks & Outdoor, Low-power MAC \\ 
    120 Deployment 3 & & 23 & 5 & One week  & Industrial environment \\ 
     119% Deployment 2 & Hydrowatch 2 \cite{hydrowatch} & 32 & 5 & Two weeks & Outdoor, Low-power MAC \\ 
     120% Deployment 3 & & 23 & 5 & One week  & Industrial environment \\ 
    121121\hline 
    122122\end{tabular} 
    … …  
    136136to TinyAODV, a compact implementation of the AODV 
    137137on-demand distance vector routing protocol \cite{aodv}.   
    138 Conceptually, the combination of CTP and AODV is very 
    139 similar to 802.11s, which is perhaps the existing architecture closest to our 
    140 design. 
     138% Conceptually, the combination of CTP and AODV is very 
     139% similar to 802.11s, which is perhaps the existing architecture closest to our 
     140% design. 
    141141 
    142142% \SDHnote{too whiny?  nobody likes a whiner.} 
    … …  
    179179  real-world deployment.  {\lowthane}'s collection performance suffers 
    180180  slightly when faced with interference on a week day 
    181   (\ref{fig:collection-acme}), but compares favorably with state-of-the-art 
     181  (\ref{fig:collection-acme}), yet still compares favorably with state-of-the-art 
    182182  CTP in testbed experiments.} 
    183183\label{fig:collection} 
    … …  
    257257exploration rate, which determines how aggressive nodes are in gathering 
    258258information about neighboring nodes\footnote{In our implementation, a 
    259     maximum of 4 mature neighbors are be reported in a topology report.  Any 
    260     new information received by a controller expires all old information.}.  In essence, the degree of a node in the global topology is determined by the number of neighbors in the node's topology report, as well as the number of other node's topology reports that this particular node appears in. 
     259    maximum of 4 mature neighbors are reported in a topology report.  Any 
     260    new information received by a controller expires all old information 
     261    from that node.}.  In essence, the degree of a node in the global topology is determined by the number of neighbors in the node's topology report, as well as the number of other node's topology reports that this particular node appears in. 
    261262% To quantify the impact of these two parameters, we evaluated {\lowthane}'s performance on Testbed A from an uninitialized state.  In each experiment, the nodes begin with an interval of 1 second between the first two topology reports, and this interval increases geometrically until it reaches the designated Topology Report Interval, which is the steady state.  Figure~\ref{fig:top-report} shows the relationship between the average node degree in the global topology graph and the time elapsed since the network was initialized for three different scenarios: topology report intervals of 30 seconds and 5 minutes (while using the default 25\% exploration probability), and a topology report interval of 30 seconds with a halved exploration rate of 12.5\%. 
    262263The key implication is that {\emph{basic connectivity is established very 

  • docs/Lowthane/ipsn10/intro.tex

    r2784 r2788  
    3333minimize state and functionality in the network.  Our design revolves around 
    3434controlling the tradeoff between state and stretch, and minimizing control 
    35 traffic through a focus on data-driven design.%   We embrace the open-standards 
     35traffic through a focus on data-driven design.  %   We embrace the open-standards 
    3636% approach to system design, and our solution, {\lowthane}, is in fact the 
    3737% default routing protocol for blip, a widely-used IPv6 network layer for 

  • docs/Lowthane/ipsn10/overview.tex

    r2784 r2788  
    3030 
    3131{\lowthane} transmits a portion of the topology discovered by the 
    32 collection protocol to higher-function {\controller}s.  An significant 
     32collection protocol to higher-function {\controller}s.  A significant 
    3333advantage of using this topology is that collection traffic, which is 
    3434often present as a background workload in this class of network, is