Scientific research journal of india (SRJI Vol-2 Issue-2 Year-2013)

Page 53

ISSN: 2277-1700 ● Website: http://www.srji.info.ms ● URL Forwarded to: http://sites.google.com/site/scientificrji

These algorithms (e.g., ERICA, ERICA+) are designed for the ATM Available Bit Rate (ABR) service and require all network switches to compute fair allocations of bandwidth among competing connections. However, these algorithms are not easily tailor able to the current Internet, because they violate the Internet design philosophy of keeping router implementations simple and pushing complexity to the edges of the network. Fig: 2, The Core-Stateless Internet Architecture

Rangarajan and Acharya proposed a network

Assumed By NBP

border-based approach, which aims to prevent

1.1 TCP congestion control has mainly two

congestion collapse through early regulation of

phases:

unresponsive flows (ERUF)12. ERUF border

Slow Start and Congestion avoidance. A

routers rate control the input traffic, while core

new connection begins in Slow-start, setting its

routers generate source quenches on packet drops

initial cwnd to 1 packet, and increasing it by 1 for

to advise sources and border routers to reduce their

every received Acknowledgement (ACK). After

sending rates. While

this

approach

may

cwnd reaches ssthresh, the connection switches to

prevent

congestion-avoidance where cwnd grows linearly.

congestion collapse, it does so after packets have

A variety of methods have been suggested in the

been dropped and the network is congested. It also

literature recently aiming to avoid multiple losses

lacks mechanisms to provide fair bandwidth

and achieve higher utilization during the startup

allocations to flows. That is responsive and

phase. A larger initial cwnd, roughly 4K bytes, is

unresponsive to congestion. Floyd and fall have

proposed in.

approached the problem of congestion collapse by

This could greatly speed up transfers with

proposing low-complexity router mechanisms that

only a few packets. However, the improvement is

promote the use of adaptive or “TCP-friendly”

still inadequate when BDP is very large and the

end-to-end congestion control10. Their suggested

file to transfer is bigger than just a few packets.

approach requires selected gateway routers to monitor determine

high-bandwidth whether

they

flows are

in

order

to

responsive

to

Fast start uses cached cwnd and ssthresh in recent connections to reduce the transfer latency. The cached parameters may be too aggressive or too

congestion. Flows determined to be unresponsive

conservative when network conditions change

to congestion are penalized by a higher packet

Smooth start has been proposed to slow down

discarding rate at the gateway router. A limitation

cwnd increase when it is close to ssthresh. The

of this approach is that the procedures currently

assumption here is that default value of ssthresh is

available to identify unresponsive flows are not

often larger than the BDP, which is no longer true

always successful5.

in large bandwidth delay networks. Proposes to set the initial ssthresh to the BDP estimated (Packet Network Discovery) has been proposed to derive 46


Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.