Th`ese de Doctorat de l'université Paris VI Pierre et Marie Curie Mlle ...
Th`ese de Doctorat de l'université Paris VI Pierre et Marie Curie Mlle ...
Th`ese de Doctorat de l'université Paris VI Pierre et Marie Curie Mlle ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
To implement these two steps, it is necessary to have an accurate information regarding<br />
all connections entering the n<strong>et</strong>work. More specifically, at the beginning of each update<br />
interval, our algorithms need to know the state of each connection and its actual sending<br />
rate.<br />
To this end we installed traffic monitors at each ingress router to perform online mea-<br />
surements on the incoming traffic flows, and we implemented a table, named Connections’<br />
State Table (CST), that contains the state information concerning all connections.<br />
The J-Sim n<strong>et</strong>work simulation package (drcl.in<strong>et</strong>) offers the basic classes <strong>de</strong>fined in the<br />
abstract n<strong>et</strong>work mo<strong>de</strong>l. Therefore, to augment the ingress router elements with the key<br />
features mentioned above, we exten<strong>de</strong>d the drcl.in<strong>et</strong> package to implement the CST table<br />
and traffic monitors.<br />
The structure of this Chapter is the following: Section A.1 <strong>de</strong>scribes in some <strong>de</strong>tail<br />
AMPL, and Section A.2 provi<strong>de</strong>s a brief <strong>de</strong>scription of our extensions to J-Sim existing<br />
classes, including the CST table and dynamic bandwidth allocation algorithms implemen-<br />
tation.<br />
A.1 AMPL: A Mo<strong>de</strong>ling Language for Mathematical<br />
Programming<br />
AMPL is an algebraic mo<strong>de</strong>ling language for linear and nonlinear optimization problems, in<br />
discr<strong>et</strong>e or continuous variables. Developed at Bell Laboratories, AMPL allows to formulate<br />
optimization mo<strong>de</strong>ls and examine solutions, while the computer manages communication<br />
with an appropriate solver. AMPL’s flexibility and convenience make it i<strong>de</strong>al for rapid<br />
prototyping and mo<strong>de</strong>l <strong>de</strong>velopment, while its speed and control options make it an efficient<br />
choice for repeated production runs. For more <strong>de</strong>tails, the rea<strong>de</strong>r can refer to [35].<br />
As <strong>de</strong>scribed in the previous Chapters, the two algorithms OBA and IBA are based<br />
on a mathematical formulation of the n<strong>et</strong>work revenue maximization problem. Therefore,<br />
94