Made available in DSpace on 2014-06-12T16:00:28Z (GMT). No. of bitstreams: 2 arquivo6669_1.pdf: 1298139 bytes, checksum: 82c0aa9def52f663c245e3f57be952ef (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2007 / Conselho Nacional de Desenvolvimento Científico e Tecnológico / The Transmission Control Protocol (TCP) is responsible for supplying reliable data transport service on the TCP/IP stack and for carrying most than 90% of all Internet traffic. In addition, the stability and efficiency of the actual TCP congestion control mechanisms have been extensively studied and are indeed well known by the networking community. However, new Internet applications and functionalities continuously modify its traffic characteristics, demanding new research in order to adapt TCP to the new reality of the Internet. In particular, a traffic phenomenon known as "mice and elephants" has been motivating important researches around the TCP. The main point is that the standard TCP congestion control mechanisms were designed for elephants leading small flows to experience poor performance. This is caused by the exponential behavior of Slow Start which often causes multiple packet losses due their aggressive increase. This work examines minutely the problems caused by the standard TCP congestion control to mice flows as well as it studies the most important proposals to solve them. Thus, based on such research studies, a modified TCP startup mechanism was proposed. The Burst TCP (B-TCP) is an intuitive TCP modification that employs a responsive congestion window growth scheme based on the current window size, to improve performance for small flows. Moreover, B-TCP is easy to implement and requires TCP adjustment at the sender side only. Simulation experiments show that B-TCP can significantly reduce both transfer times and packet losses for small flows without causing damage to large flows