Skip to main content

Table 4 Comparison of the main characteristic of various TCP Enhancements

From: TCP performance in multi-hop wireless ad hoc networks: challenges and solution

 

"FeW"

"APS-FeW"

"Vegas-W"

"Link RED and AP"

"CWL"

 

[48]

[49]

[47]

[46]

[24]

DEALING WITH ROUTING FAILURES

No.

No.

No.

No.

No.

DEALING WITH WIRELESS ERRORS

No.

No.

No.

No.

No.

DEALING WITH CONTENTION

Yes: it reduces the TCP window over-shoot to reduce the contention.

Yes: it reduces the TCP window over-shoot to reduce the contention.

Yes: can be reduced by controlling the rate.

Yes: it detects early sign of congestion. The sending rate is reduced to improve spatial reuse and alleviate the contention.

Yes: contention can be reduced by limiting the cwnd.

DEALING WITH RETRANSMISSION

No.

No.

No.

No.

Yes: the maximum retransmission time-out of TCP is modified to 2 second instead of 240 second.

DEALING WITH TCP CONGESTION CONTROL MECHANISM

Yes: the cwnd is updated by fractional window.

Yes: it uses byte count instead of packet count to calculates cwnd. The packet size varies based on cwnd size.

Yes: both slow start and congestion avoidance mechanisms in TCP-Vegas are modified to increase cwnd after receiving more than one ACK. It also updates slow start threshold by tracking stable window.

No.

Yes: limiting the congestion window by the round-trip-hop-count.

DEALING WITH HIDDEN EXPOSED PROBLEM

No.

No.

No.

Not directly but the problem can be reduced by considering spatial channel reuse.

Not directly but the problem can be reduced by considering spatial channel reuse.

DEALING WITH TCP RATE

Yes: low rate caused by small window.

Yes: low rate caused by small window.

Yes: slow rate cased by congestion window fractions and rate control timer.

Yes: slow rate caused by (ECN) marking.

Yes: low rate caused by limited congestion window.

DEALING WITH TCP ACK

No.

No.

No.

No.

No.