Paper: 298637
Title: Neural Network Congestion Control Design for TCP/IP Network Based Genetic Algorithm

-------------------- review 1 --------------------
---------------------------- REVIEW 1 --------------------------
PAPER: 2
TITLE: Neural Network Congestion Control Design for TCP/IP Network Based Genetic Algorithm
 
OVERALL RATING: 2 (accept) 
REVIEWER'S CONFIDENCE: 2 (medium) 
Relevance to this conference: 4 (good)
Originality/Uniqueness: 4 (good)
English readability: 3 (fair)
Paper organization/presentation: 3 (fair)
Has good survey been done?: 3 (fair)


COMMENTS:
1. On page 3, First equation should be PI, should not it.
2. The authors describe the use of a dynamic recurrent neural model.
However, Fig. 5 shows only a feedforward network. I wonder where the recurrent
neural network involves in the controller.
3. Although, the authors describe the use of GA for changing weights of the network,
it is still unclear how the weights of the network are really updated. 
For clarity, I would recommend to also present the equations of weight changes. 



-------------------- review 2 --------------------
---------------------------- REVIEW 2 --------------------------
PAPER: 2
TITLE: Neural Network Congestion Control Design for TCP/IP Network Based Genetic Algorithm
 
OVERALL RATING: 2 (accept) 
REVIEWER'S CONFIDENCE: 3 (high) 
Relevance to this conference: 4 (good)
Originality/Uniqueness: 4 (good)
English readability: 4 (good)
Paper organization/presentation: 5 (excellent)
Has good survey been done?: 4 (good)


Authors introduce a novel AQM methodology   to overcome problems that are identifying to TCP networks dynamics due.  Methodology can improve the type of controller design to previous controller design (PI, RED, PID and APC).