Transmission control protocol or (TCP) is used purposely for data transfer over the internet. The main strength on this protocol is its ability to make use of flow control algorithm which allows TCP to decrease the rate at which data is sent when congestion in the network occurs. For this particular paper, research will be carried out to find out how TCP can transmit data for wireless networks.
This research is carried out since TCP was originally designed for networks that had wired links and carried stationary hosts. However, there has been an increase in the use of wireless networks, which present a whole new set of diverse characteristics different from the wired networks. This is because wireless networks are characterized by wireless links and hosts that are mobile with a very bit error rate of BER 10-2 to 10-4 as compared to the fiber link of 10-9 (Chan, Tsang & Gupta, 1997). These high bit error rates have been known to cause, packet errors that cause false alarms and which, cause unnecessary congestions, back-off in retransmission time at the host. This error is also associated with mobility and signal fading, associated with non-congestive and congestive losses as defined by (Leung & Li, 2006). Therefore, this research seeks to find current information on how TCP is calibrated and enhanced to address the problem of congestion and packet losses for wireless networks. This is based on the fact that current communication systems are being faced by wireless applications like high speed multimedia and wireless communication platforms carried by the internet (Tian, Xu & Ansari, 2005).
The Transmission Control Protocol for the internet was created by American department of defense, under the body, Advanced Research Projects Agency Network (ARPANET) (Leung & Li, 2006). ARPANET made use of the packet switching technology to send data, and consequently under Request for Comment or RFC 793, TCP was created, while RFC 791 created IP in 1981, by the Internet Engineering Task force of IEFT (Tian, Xu & Ansari, 2005). APRANET built TCP in the many layers of the open system interconnection or the OSI. TCP was created as a protocol in the layer four transport systems, which uses the fundamental IP services to give applications on a connection that is end-to-end, for data delivery (Tian, Xu & Ansari, 2005).
Originally, ARPANET created TCP purposely for wired networks. In these networks, there is negligible BER and congestion which were cited as the major causes of packet loss (Tian, Xu & Ansari, 2004). This protocol implements packet flow through flow and congestion control algorithms which are founded on additive multiplicative decrease or (AIMD) and sliding window algorithms (Tian, Xu & Ansari, 2004). The purpose of the sliding window is to offer a control mechanism which allows a sender to send a transmission window when they receive an acknowledgement (ACK) which indicates the last packet which is in in-order, as received successfully by the receiver (Tian, Xu & Ansari, 2004). In cases, where packets are lost due to congestion caused by buffer overflow, either the sender will receive a duplicate ACKs or they will retransmit timeout (RTO). These actions cause the sender to reduce their transmission or congestion window (cwnd) (Tian, Xu & Ansari, 2004). To eliminate the failure of packet loss, TCP was designed to implement flow and congestion control algorithms founded on the additive increase multiplicative decrease and the sliding window.
For TCP, fast recovery works through the comparison of fast retransmit. Fast transmission suggests that there is a presence of mild network congestion. This implies that here is need to set ssthresh to half of the data sent that is outstanding in the network (Leung & Li, 2006). By receiving the duplicate ACK, this indicates a departure from the network by the segment, and cwnd is then set the total ssthresh and the amount of duplicate ACKs received (Leung & Li, 2006). During transmission, when an acknowledgement or ACK for new segment arrives, cwnd is reset to ssthresh and a triggering of congestion avoidance occurs (Leung & Li, 2006). Of importance to Transmission Control Protocol in the transfer of information or data over the internet are the variations, TCP Tahoe and /TCP Reno (Leung & Li, 2006). These are different in terms of, while TCP Tahoe has slow start, fast transmit and congestion avoidance, TCP Reno adds more fast recovery to congestion control mechanisms levels of TCP Tahoe (Leung & Li, 2006).
In order for TCP, to operate in the wireless networks several changes had to be made to it. TCP performance in a wireless environment can be improved by distinguishing corruption from congestion or by the improvement of the quality of TCP connections through partial error recovery on the layer underneath TCP a suggested by (Chan, Tsang & Gupta, 1997). Any design changes to TCP must factor in the main characteristics of wireless networks that make them distinctively different from wired networks. TCP must be designed to meet the demands of wirelesses channel contention, mobility, signal fading and limited energy and power characteristics (Leung & Li, 2006).
Since wireless networks are being employed widely in today’s communication field, several changes have been made to TCP. These changes are based on the type of wireless network used by considering its characteristics and have been effectively utilized by Google in its operations. The most common are satellite, ad hoc, cellular and wireless LANs. TCP schemes designed for satellite networks are founded on the fact that the sender takes a long time to get to the high sending rate during a slow start phase (Tian, Xu & Ansari, 2005). Therefore, for this network, new TCP-peach make use of rapid recovery and sudden start to overcome the problems of satellite communication. For ad hoc wireless networks, ATCP is designed as an improvement of the end-to-end solution, by the implementation of a thin layer which is inserted between the IP and standard TCP layers (Tian, Xu & Ansari, 2005). This design relies on ECN or explicit congestion notification to detect congestion. For cellular wireless networks, modifications are based on the handoff problems of cellular networks and their high BER (Tian, Xu & Ansari, 2005). Freeze-TCP is an end-to-end solution designed for cellular wireless networks, as it imposes on the restrictions of routers and only uses code modification for mobile unit (Tian, Xu & Ansari, 2005). This sets the advertiser window to zero on the ACK packets in the presence of disconnection.
TCP experiences poor performance due to packer re-ordering and non-congestive segment loss like burst and random losses. Random loss entails he sudden loss of data, due to fading signals in the wireless network. This is caused when the congestion control mechanisms for TCP react inappropriately by keeping the send rate for connections small and by retransmitting some segments spuriously (Leung & Li, 2006). Bust losses occur when signal fading occurs, or when thee is a very long uncontrolled interference in the channel (Leung & Li, 2006). Burst losses are unlike random losses, since they happen over a very short time duration which causes consecutive segments to be lost. All burst loss events must occur if the transmission path is interrupted for whatever reason (Leung & Li, 2006). Lastly, TCP faces the problem of packer re-ordering which involves how data packets are received by the receiver as compared to what was sent by the sender. The other major problem with data transmission over the internet is the difficulty of separating TCP for data transfer that is utilizing a network that has both wired and wireless facilities. Therefore, further research is needed in addressing a TCP protocol that can effectively eliminate these problems for all wireless networks.
Chan, A.C.F., Tsang, D.H.K. and Gupta, S. (1997). “TCP (Transmission Control Protocol) over wireless links.” Vehicular technology conference, IEEE 47th 3, 1326-1330.
Ka-Cheong Leung and Li, V.O.K. (2006). “Transmission control Protocol (TCP) in wireless networks: Issues, Approaches, and Challenges.” Communications surveys & Tutorials, IEEE, 8.4, 64-79.
Qureshi, M.J.A. and Saleem, M. (2007). “Simulation and Visulization of Transmission Control Protocol’s (TCP) flow-Control and Multi-Home options”. Applied Sciences & Technology, IBCAST, IEEE 8-11 Jan. 2007, 139-146.
Tian Ye, Xu Kai and Ansari, N. (2005). “TCP in wireless environments: problems and solutions.” Communication magazine, IEEE 43.3, s27-s32.
Tian Ye, Xu Kai, and Ansari N. (2004). “TCP-jersey for wireless IP communications.” Selected areas in communications, IEEE journal 22.4, 747-756.