TCP for Wireless Environments

International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-2, Issue-11, Nov- 2016]
Infogain Publication (Infogainpublication.com) ISSN : 2454-1311
www.ijaems.com Page | 1905
TCP for Wireless Environments
Mr. Ranjeet V. Bidwe, Mr. Mahesh A. Kandekar
Dept of Computer engineering, PICT, Pune. Maharashtra, India.
Dept of Computer engineering, AISSMS COE, Pune. Maharashtra, India.
Abstract— Computer networks have experienced an explosive
growth over the past few years, which has lead to some severe
congestion problems. Reliable protocols like TCP works well
in wired networks where loss occurs mostly because of
congestion. However, in wireless networks, loss occurs
because of bit rates and handoffs too. TCP responds all losses
by congestion control and avoidance algorithms, which
results in degradation of TCP’s End-To-End performance in
wireless networks. This paper discusses different issues and
problems regarding use of TCP in wireless networks and
provides comprehensive survey of various schemes to improve
performance of TCP in Wireless Networks.
Keywords—TCP, Mobile-IP, Wireless networks, Protocol
design.
I. INTRODUCTION
Due to rapid advances in the area of wireless communications
and the popularity of the Internet, the provision of packet data
services for applications like e-mail, web browsing, mobile
computing etc. over wireless is gaining importance. The
TCP/IP protocol suites have number of layers, of which
transport layer is used widely for mobility. It uses protocols
like TCP and UDP for transferring data.
While coming towards wireless environment, we must
understand what wireless environment is first. Wireless
environment can be broadly distinguished in three types:
Cellular networks, Ad-hoc networks and Satellite networks. In
Cellular Networks a mobile host is connected to the fixed
network with the help of the Base Station. This is the most
common form of Wireless Network currently in use. Mobile
devices like cell phones, laptops use this network. Most of the
proposed solutions to TCP use this model. All service
providers are on the fixed network and hence we have to
address the problem of wireless networks only at one end
point. Ad – Hoc Networks are formed by mobile hosts which
are connected to each other within a radio distance. This kind
of a model is not well deployed and very few solutions have
been proposed to this model. Satellite Networks are those
where a satellite link is in between the sender and the receiver.
These have very high BERs (Bit Error Rates) and high latency
because the Satellite are at a great distance from the surface.
Now going towards our main concept, Transmission Control
Protocol (TCP) [1] is one of the important standards in the
internet world and also a very vital element in internet
protocol suite. It provides a connection oriented service with
reliable data transfer over the unreliable underlying protocols.
It uses sequence numbering and timers to ensure reliable
transfer of packets. TCP's flow control increases the data
sending rate until there are signs of congestion in the network.
The basis of TCP congestion control lies in the following
algorithms: slow start, congestion, avoidance, fast retransmit
and fast recovery [2].
In the following we first outline the different issues of TCP in
wireless networks. Then the main problem regarding TCP.
Then we summarize some proposed solutions with their
strengths and weaknesses.
II. ISSUES IN WIRELESS ENVIRONMENTS
Wired services are relatively reliable compared to wireless
networks. So if any packet get lost then it is due to congestion
only, so that they can carried out a congestion control scheme
to get lost packets. But in wireless networks some serious
issues are found, those are:
First issue is Bit Error Rate (BER). Wireless host uses radio
transmission or infrared wave transmission for
communication. Experimentally found that, The BER of
wireless links is typically higher than that of wired networks.
Also BER also varies by a large amount when wireless
environment changes quickly.
Second issue is Bandwidth. Wireless links having very less
bandwidth as compared to the wired links. Wireless links
offers bandwidth of 2MBPS, while wired links offers 10-100
MBPS. As wireless links offers very low bandwidth, Optimum
use of available bandwidth is a major issue in heterogeneous
networks that has to be taken care of.
Third issue is Mobility. As world is moving towards wireless
environment, large addition of mobile devices are done. So it
introduces huge amount of indeterminate mobility in rather a
stationary network. This tends to introduce some amount of
instability in existing network topology. When wireless host is
moving in a particular network, its base station is sending data
to it. But when it moves to another station during handoff, the
data sent by old base station is lost as it moved out of range.
Similarly data it is sending to old base station is lost.
Next issue is Round Trip Time (RTT). The wireless media
exhibits longer latencies than wired media in the case of
satellite networks. It is almost the same as in wired networks
since Radio waves travel at the speed of light which is same as

the transmission speeds in wired media. Since the bandwidth
is lower in wireless networks a packet takes longer to get
transmitted in wireless networks. This affects overall
throughput and increases interactive delays.
Last issue is Power consumption. Normally mobile hosts have
limited power and processing speed compared to base stations,
which forms inefficiency in network. Solutions that Take
power consumption into account have a clear-cut advantage
over the otherwise designed solutions.
TCP works reliably well on wired networks and fixed
topologies, so it operates on assumption that packet lost is due
to congestion. But this assumption is not true in case of
wireless networks. There are many reasons of packet loss like
disconnection, corruption by underlying physical medium,
handoffs, but TCP assumes it as due to congestion in network.
So it cannot find actual reason behind loss of data. But this
wrong assumption degrades the TCP performance. For
example, let data is lost due to temporary or short
disconnection, but TCP assumes it is due to congestion and
decrease the window size to minimum size, and starting the
slow start mechanism [2] which means that sender
unnecessarily holds back, slowly growing the transmission
rate. Even though receiver recovers quickly from temporary or
short disconnection. This is illustrated in following Fig.1
where it is seen that the network capacity can remain
unutilized for a while even after a reconnection.
Fig.1: TCP SLOW START [10]
The fundamental problem is the underestimation of bandwidth
by the network endpoints which results in reduced application
layer performance, reductions in throughput and unacceptable
delays. When this happens the applications don’t get their fair
share of the bottleneck link’s bandwidth.
Another problem may happen during handoff, there are three
major impacts on TCP during the handoff scenario. The
packets will experience a higher delay during handoff due to
packet re-routing. Secondly the packets already in transit for
the old access point are generally dropped during the handoffs.
Lastly TCP has to deal with massive packet re-ordering after a
handoff. Same issues are discussed in various papers. [21]
III. DIFFERENT APPROACHES IN WIRELESS
NETWORKS
In this section, we are discussing some approaches proposed
by taking problems under consideration to improve the
performance of TCP over wireless environment.
Snoop [3],[4]protocol is classified as a TCP Aware link-level
protocol. In this protocol a network layer software is updated
at Base station (BS) by adding module called snoop. Snoop
module checks every packet travelling on the connection in
both directions. It maintains cache of TCP packets which are
sent by fixed host (FH) to mobile host (MH) but not
acknowledged by MH. When packet is sent from FH, snoop
adds it to cache and forwards it according to its routing
information. It also checks acknowledgment coming from
MH, if any packet gets lost or snoop got any duplicate
acknowledgment about packet, and then it resends that packet
if it is cached. It maintains its own timers for retransmission of
buffered packets, implements selective transmission etc. by
this way snoop hides loss of packets from FH, by not
propagating duplicate acknowledgments, and thereby it
prevents further invocations of congestion control mechanism
[5].
Main disadvantage of this scheme is that, it relies on
intermediaries i.e.. BS, so it does not satisfies true end to end
semantic proposed by TCP. This protocol does not completely
shield the sender from wireless losses as the sender may
timeout due to repeated losses or bit errors caused by the
wireless link.
An extension proposed to random delay detection (RED) is
Explicit Congestion Notification (ECN) [5]. RED is an active
queue management mechanism in routers, it detects
congestion before the queue overflows and provides an
indication of this congestion to the end nodes. A RED router
signals incipient congestion to TCP by dropping packets
before the queue runs out of buffer space. RED router operates
by maintaining two levels of thresholds minimum ( minth )
and maximum ( maxth ). If the average queue size lies
between the minth and maxth , then It drops packets. ECN
is extension to RED, which marks a packet instead of
dropping in when the average queue size lies between minth
and maxth . Upon receipt of congestion marked packet, the
TCP receiver informs the sender (by subsequent
acknowledgement) about happening congestion, which starts
the congestion avoidance algorithm at the sender. ECN
requires support from both the router as well as the end hosts,
there is need of modification at the end host of TCP stack. If
the ECN support is provided then the packets are referred as
ECN capable packets. RED droops packets that are not ECN
capable.
Explicit Bad State Notification (ESBN) [6] proposes a
mechanism to update the TCP timer at the source to prevent
source from decreasing its congestion window, if there is
congestion occurring. EBSN’s are sent to the source, when
base station is trying to send a packet over wireless link and
fails to send. EBSN would cause the previous timeouts to be

cancelled and new timeouts put in place, based on existing
estimate of round trip time and variance. Thus, the new
timeout value is identical to the previous one. The EBSN
approach does not interfere with actual round trip time or
variance estimates and at the same time prevents unnecessary
timeouts from occurring. This prevents timeouts for packets
that had already been put on the network before the wireless
link encountered the bad state.
Explicit Loss Notification (ELN) [7] adds an ELN option to
TCP acknowledgment. When a packet is dropped on the
wireless networks, future cumulative acknowledgements
corresponding to the lost packet are marked to identify that a
non-congestion related loss has occurred. Upon receiving this
information along with duplicate acknowledgements, then
sender may retransmit data instead of congestion control
algorithms.
Holland and Vaidya proposed a feedback based technique
called TCP-ELFN [8][9]. ELFN stands for Explicit Link
Failure Notification. The goal is to inform TCP sender of link
and route failures so it can avoid responding to the failures as
if congestion occurs. ELFN is based on DSR[10] routing
protocol. To implement ELFN message, the route failure
message of DSR is modify to carry payload, it is similar to
“host unreachable” ICMP message. Upon receiving ELFN
message, TCP sender disables congestion control mechanism
and enters in stand-by mode, it sends a small packet to probe
the network to see if route has been established. If new route
has been established, then it leaves stand-by mode, restores its
retransmission timer (RTO) and continues as normal. Though
explicit route failure notification, TCP-EFLN allows sender to
instantly enter in stand-by mode to avoid unnecessary
transmission and congestion control, which wastes precious
MH battery power and scarce bandwidth. With explicit route
reestablishment notification from intermediate nodes or active
route probing initiated at the sender, these two schemes enable
the sender to resume fast transmission as soon as possible. But
neither of these two considers the effects of congestion, out-
of-order packets, or bit errors, which are quite common in
wireless ad hoc networks.
Next approach I-TCP [11] splits connection between FH and
MH in two parts. First part is FH to BS and from BS to MH.
Firstly FH sends data to BS, BS acknowledged that data, then
it is responsibility of BS to forward that data to MH. This
indirection helps shield the wired network from the
uncertainties of the wireless network and the TCP/IP at the
fixed host side need not be changed. On the link between BS
and MH, it is not necessary to use TCP. One can use any other
protocol optimized for wireless links.
Using indirection in this method tends to number of benefits.
It separates flow control and congestion control functionality
on the wireless link from that on the fixed network. Also
Indirection allows the BS to manage much of the
communication overhead for a mobile host. I-TCP also
optimizes the handoff by shrinking the receive window size at
the MSR which forces the FH to stop sending data when the
MSR buffers are full. Drawback to this protocol is again it is
not following true end-to-end semantics of TCP. In case of
transferring data, copying data from the incoming connection
from the FH to the outgoing connection to the MH is also
needed. In case of frequent handoffs, the overhead related to
the connection state transfer between the base stations may be
large and add delays. And also the base stations have to be
complex and with large buffers in case of heavy traffic.
Working of I-TCP is case of handoff to transfer connection is
shown in Fig.2
Next approach is MTCP [11]. MTCP is similar to I-TCP and
also splits a TCP connection into two: one from MH to BS and
the other from BS to FH. The MH to BS connection passes
through a
Fig.2: I-TCP Connection Transfer during Handoff [11]
Session layer protocol which can employ a selective repeat
protocol (SRP) over the wireless link.Most of the schemes
proposed for optimizing Transport Layer (TCP) over wireless
networks needs intermediaries, due to which the end to end
semantics of TCP are not maintained and problems like
degradation in throughput are resulting. So next proposed
protocol i.e. Freeze-TCP [13] satisfies true end to end
semantics of TCP, which does not require any intermediaries;
neither change in TCP code is required on the sender side or
the intermediate routers. Change is limited to the mobile client
side, and hence is interoperable with the existing networks.
In Freeze-TCP, receiver identifies an impending disconnection
because of potential handoff, fading signal strength, or any
other problem arising due to wireless media and notifies the
sender of any impending disconnection by advertising a zero
window size (ZWA- zero window advertisement) and prevents

the sender from entering into congestion avoidance phase.
Upon getting the advertised window as zero, the sender enters
the persist mode and freezes all the timers related to the
session. And periodically sends the ZWP (Zero Window
Probes) until the receiver’s window opens up. Since the ZWPs
are exponentially backed off, there is a possibility of having a
long idle time after the reestablishment of connection. To
avoid this, the receiver employs “TR-ACKs” (Triplicate
Reconnection ACKs). As soon as the connection is
reestablished, the receiver sends 3 copies of the ACK for the
last data segment successfully received prior to disconnection
to enable the fast transmit.
Fig.3: Increased throughput due to Freeze-TCP [13]
But main dis-advantage of Freeze-TCP is that, Freeze-TCP is
only useful, if a disconnection occurs while the data is being
transferred. It is not useful, in case of a disconnection when no
data is being transferred between sender and receiver.
Another approach where we preserve end to end semantics is
WTCP [14]. It was developed for Wireless Wide Area
Networks (WWAN) where the TCP algorithms failed because
it falsely assumes packet losses are due to congestion. This
protocol distinguishes congestion losses and random losses. It
uses packet departure time and packet arrival time for that.
WTCP shapes traffic since it uses rate based transmission
control, it never allows burst of packet transmission. This is
useful when different connections have different Round Trip
Times (RTT). The basic idea behind this protocol is that TCP
should not half its transmission rate for just a packet loss
which happens more frequently in wireless Networks. This is
more like an algorithm where the receiver takes the
responsibility of receiving all packets. The sender does not
decide which packets have to be transmitted because some of
the ACKs have failed but they probe the receiver to find out if
a packet has to be resent. WTCP uses the ratio of the inter-
packet separation at the receiver and the inter-packet
separation at the sender as a metric for rate control rather than
using packet losses and retransmit timeouts. WTCP reuses the
standard TCP mechanism for flow control and connection
management. It uses inter packet delay as a metric for
congestion control, using this it performs the rate adaptation
computation at receivers end. Also it provides fairly accurate
measure of the available channel rate for low bandwidth
channels.
However just accepting small losses as random may cause it to
disregard incipient congestion. WTCP thus maintains a history
of losses and reduces transmission rates more aggressively if
they happened quickly. Since it does not use ACKs as a metric
even Startup transmission is measured by inter-packet delay.
Thus at startup WTCP sends a packet pair and uses that to
adjust to the network behavior.
Next proposed approach is TCP Santa Cruz [15]. This
protocol also uses same approach as WTCP. TCP Santa Cruz
monitors the queue developing over a bottleneck link and this
determines whether congestion is increasing in the network.
Using this it identifies the type of loss, may be congestion or
random and it responds it appropriately. It is able to find out
direction of congestion with initial stage of congestion.
Congestion is determined by calculating the relative delay that
one packet experiences with respect to another as it traverses
the network.
It is observed that losses due to congestion are followed by an
increase in the network bottleneck queue. A wireless loss on
the other hand, can be identified as a random loss that is not
followed by a build-up in the bottleneck queue. TCP-SC
monitors changes in the bottleneck queue over an interval
equal to the amount of time it takes to transmit one window of
data and receive acknowledgements corresponding to all the
packets transmitted in the window. When these losses are
discovered, then we expect the protocol to simply retransmit
most losses without affecting the transmission window. This
can be implemented as a TCP option by utilizing the extra 40
bytes available in the options field of the TCP header.
Next approach M-TCP [16] works well in frequent
disconnection and low bit rate wireless links. The spurious
time out as shown in the Fig.4 below proves to be very
harmful to overall throughput than losses due to but errors or
small congestion windows.
Fig.4: Serial timeouts at TCP sender [12]
In M-TCP, every TCP connection is split in two parts at the
Supervisory Host (SH) . TCP connection from fixed host (FH)
to the SH uses the standard, unmodified version of TCP. And
connection between SH and mobile host (MH) uses the
modified version of TCP. Wireless bandwidth is important
resource here, and it should be keenly used. In heterogeneous
systems, there is variation in available bandwidth. But SH
takes care of it.
Firstly FH sends segment, then it is taken by SH. Then SH
forwards that segment to MH. Then MH gives
acknowledgment for that segment.SH, upon getting acks,

acknowledges back to FH. Unlike other split connection
techniques, it saves the ack of the last byte, in order to prevent
loss of outstanding packets. Now in case, if the MH is
disconnected from nowhere, then the SH stops getting the acks
and assumes that MH has been temporarily disconnected and
sends the ack of the last byte that it saved previously. This ack
will contain the advertised window of the MH as “zero”, then
sender enters the persist mode and freezes all the timers
related to the session, and starts sending the exponentially
backed off persist packets to the SH. The SH responds with
the zero window size at the receiver, to each persist packet,
until it receives some nonzero window size indication from the
receiver. When it receives, then SH immediately replies to the
persist packet as the appropriate window size and resumes all
its freezed timers. Thus the sender can resume transmitting at
full-speed. The FH again starts transmitting from the next byte
that is unacknowledged.
The state transition diagram for ATCP at the sender is shown
in Fig. 6. Upon receiving a “Destination Unreachable”
message, the sender enters the persist state. The TCP at the
sender is frozen and no packets are sent until a new route is
found, so the sender does not invoke congestion control. Upon
receipt of an ECN, congestion control is invoked without
waiting for a timeout event. If a packet loss happens and the
ECN flag is not set, ATCP assumes the loss is due to bit errors
and simply retransmits the lost packet. In case of Multi-path
routing, upon receipt of duplicate ACKs,
Fig.5 Architecture for M-TCP[12]
TCP sender does not invoke congestion control, because multi-path routing shuffles the order in which segments are received.
So ATCP works well when the multi-path routing is applied.
Fig.6: State transition diagram of ATCP at sender[17]
Next approach is TCP Westwood[18]. It is a sender-side
modification of the TCP congestion window algorithm that
improves upon the performance of TCP Reno in wired as well
as wireless networks. General idea used here is to use
bandwidth estimate (BWE) to set the congestion window
(cwin) and slow start threshold (ssthresh) after congestion
episode. The main difference between TCP Reno and TCP
Westwood is, TCP Reno halves the congestion window after

three acknowledgments where as TCP Westwood attempts to
select a slow start threshold and a congestion window which
are consistent with the effective bandwidth used at the time
congestion is experience.
The source performs end-to-end estimate of the bandwidth
available along a TCP connection by measuring and averaging
the rate of returning ACKs. Whenever a sender perceives a
packet loss (i.e. a timeout occurs or 3 duplicate ACKs are
received), the sender uses the bandwidth estimate to properly
set the congestion window (cwin) and the low start threshold
(ssthresh). This way TCP Westwood avoids overly
conservative reduction of cwin and ssthresh; and thus it
ensures faster recovery.
TCP Westwood satisfies true end to end semantic of TCP.
Also it works well in mixed wired and wireless networks.
Better throughput, goodput and delay performance, fairness as
well as friendliness when coexisting with TCP Reno were
observed in experimental studies. TCP Westwood does not
require inspection and/or interception of TCP packets at
intermediate (proxy) nodes and complies with the end-to-end
TCP design principles. Only disadvantage is that it performs
poorly when random packet loss rate exceeds a few percent.
TCP-F[19] is specially designed for Ad-hoc networks. All
previously proposed schemes depend on the base station and
so cannot be applied to the multihop wireless networks, since
there are no base stations in such a network. At a time of large
data transfer from one MH to another MH through number of
MH’s, if an intermediate MH detects a route failure, due to
which it cannot send the data any further, then it sends a route
failure notification (RFN) to the source. Each intermediate
router that receives the RFN, invalidates all packets traveling
through that failed route and prevents more incoming packets.
The intermediate node than tries to find an alternate route for
the destination. If any alternate path exists, then packets are
routed through that path, otherwise RFN is forwarded towards
the source. Upon receiving RFN, source goes into the snooze
state and remains until it is notified of any updates. That time
source stops all packet sending, Marks all existing timers as
invalid, Freezes the send window of packets, Freezes value of
other state variables such as retransmission timer value and
window size and Starts a route failure timer which
corresponds to a worst case route reestablishment time. If any
intermediate router knows about a new route to the
destination, then it sends a route reestablishment notification
(RRN) packet to the source, whose identity was previously
stored. As soon as the source receives the RRN, it comes to an
active state from the snooze state. Since almost all packets in
transit would have been affected by the failure, the source
flushes out all unacked packets in its current window.
Communication would then resume at the same rate prior to
the route failure.
The TCP-Bus algorithm [20] is very similar to the TCP-F
algorithm. The basic idea is to use buffering capacity of
mobile nodes. It uses a source-initiated on-demand routing
protocol for the underlying layer. It uses two control messages
(ERDN and ERSN) related to route maintenance to notify the
source of route failures and route re-establishments. These
indicators are used to distinguish between network congestion
and route failure as a result of node-movement.
ERDN (Explicit Route Disconnection Notification) message is
generated at an intermediate node upon detection of a route
disconnection. When the sender receives the ERDN message it
stops transmitting. Similarly after discovering a new path from
the node that initiated the ERDN message the sender is
informed by using a ERSN message (Explicit Route
Successful Notification). On receiving the ERSN message the
source starts retransmission. However the retransmission of
lost packets due to congestion relies on timeout mechanism.
Since it increases the timeout to avoid retransmission during a
disconnection it must also request the lost packets as they will
be retransmitted only late. The packets from the node that
initiated the ERDN message, to the point where the node
previously existed are flushed after receiving the ERSN
message.
Fig.7: The TCP-F State Machine [19]

Hence to avoid further packet flows to the mobile node all
nodes that received the ERDN message for a particular
destination must stop forwarding those packets. Also to ensure
that an ERSN message is successfully delivered to the source,
all intermediate nodes must time out and retransmit the ERSN
message if they do not hear the upstream nodes forwarding the
ERSN message.
TCP-P [22] stands for TCP-Performance. Talking about basic
TCP functionalities, then yes, TCP-P satisfies true End to-End
semantics of TCP since no intermediaries involved. It provides
reliable, connection oriented service for mobile nodes. TCP-P
uses the standard TCP mechanisms for flow control and
connection management. Mainly TCP-P tries to solve three
important issues of TCP that are Congestion. So, TCP-P
Disconnection and Random Packet Losses mainly having
three functionalities. Working with these three functionalities
TCP-P is able to detect packet losses due to congestion in
network, disconnection in network links and random lost
packets. TCP-P is more successful than other TCP versions
since it is having more Packet Delivery Ratio as well as able
to solve more issues [23].
First function of TCP-P deals with issues like mobility and
handoff, Disconnections. It is a receiver modifying stage, here
receiver senses wireless medium continuously for detecting
fading signals which in turn detects happening disconnection.
In certain cases, it might even be able to predict a temporary
disconnection (signal strength is fading for instance). In such a
case, it advertises a zero window size, then it forces sender
into the ZWP mode and prevent it from dropping its
congestion window. When the receiver senses an impending
disconnection, first it advertises its window size as zero and a
zero window acknowledgement (called as ZWA) to sender
prior to disconnection to inform sender about disconnection.
This period is called as “warning period” (provided that
warning period should be long enough than time required for
one ZWA to get across sender). If warning period is any
longer, then sender is forced into ZWP mode. If warning
period is small then receiver will not have enough time to
inform sender, and sender have to drop its congestion window.
When connection is established again then receiver sends
three ACKs for last received packet and sender starts sending
data again. To check connection is established or not sender
sends zero window probes to receiver after an interval of time.
When data sending is going on at the same time, sender is
continuously computing the connection Bandwidth Estimate
(BWE) which is equal to the rate at which data is delivered to
the TCP receiver. The BWE value is computed by performing
end-to-end estimate of the bandwidth available along with the
TCP connection by measuring and averaging the rate of
returning ACKs. This estimated BWE value is used to set
congestion window (cwin) and slow start threshold (ssthresh)
before congestion episode. Whenever sender perceives a
packet loss (i.e. a timeout occurs or 3 duplicate ACKs are
received), the sender uses the BWE to properly set the
congestion window (cwin) and the slow start threshold
(ssthresh) and sends data accordingly. This mechanism very
little bit different from slow start mechanism. In slow start
mechanism, packet sending rate is increased exponentially i.e.
if 2 packets delivered successfully, then it will try for 4
packets, then try for 8 and so on. But in this approach packet
sending rate is not increased exponentially i.e. if 2 packets
delivered successfully, but it is not able to delivered 4 packets,
can only delivered 3 packets then it will try for sending 3
packets only, not 4 packets. It will prevent system from loss of
packets.
Initially TCP was designed with the notion in mind that wired
networks are generally reliable and any segment loss in a
transmission is due to congestion in the network rather than an
unreliable medium (The assumptions is that the packet loss
caused by damage is much less than 1%) . This notion doesn’t
hold in wireless parts of the network. Wireless links are highly
unreliable and they lose segments all the time due to a number
of factors. According to [24], noise in network is main reason
behind randomly lost packets. Up to 30% of messages can be
lost because of noise. For randomly lost packets TCP-P also
provides the solution. TCP-P just modifies the header part of
the packet. When packet is lost, i.e. its lifetime exceeds TTL
value of packet, and then lost packet itself sends loss
notification message to sender. To gain this functionality we
can modify TTL field in TCP header to send a ICMP message
to sender. This message can use value from sender IP address
from header part of TCP to send ICMP message. By this way
sender can detect lost packet and resends same packet again to
receiver.
IV. DISSCUSION
By studying above approaches, we observe that an ideal
solution should have following characteristics.
It should maintain true end to end approach without involving
any intermediaries. When we are coming towards network
security, encryption is adopted widely. whereby the whole IP
payload is encrypted, and the intermediate may not know
about the transport layer protocol used. In briefly, it should be
able to handle encryption.
Any scheme proposed must be interoperable with the existing
network infrastructure. There should be no change required in
the sender or the intermediate routers.
If at all there is any intermediate node involved in any scheme,
care should be taken of its 100% efficiency, since the
processing overheads involved with those nodes may add to
the original problem. The processing overheads may include
extra buffer space or transfer of complete state information of
a mobile node from one base station to another.
In ideal solution code at the sender should be affected. Means

we should have a static code at sender.
Also the solution should be robust against high BER. It should
be robust against frequent disconnections. And its
performance should not degrade with long disconnections.
V. CONCLUSION
In this paper, we present a comprehensive survey of the
various schemes proposed in the literature that try to solve this
problem, classified them according to their characteristics and
mentioned their limitations.
We conclude that different schemes have their own
advantages and disadvantages. But it seems that a combination
of pure link level and end to end scheme is a good
combination to alleviate the problem. Further research is
needed to investigate other approaches to help TCP
discriminate between host mobility and network congestion.
Although most schemes would yield improvement in
throughput, the key factor will be the ease with which the
modification can be incorporated in the existing infrastructure.
REFERENCES
[1] W. Richards Stevens, “TCP/IP illustrated, Volume I; The
protocols.” AWL 1994.
[2] V, Jacobson, “congestion avoidance & control”, ACM
proceedings. SIGCOMM 88.
[3] H. Balakrishan, V. N. Padmanabhan and R. Katz
“Improving reliable transport and handoff performance in
cellular wireless networks.”, wireless networks, vol.1 ,
no.4 Dec 1995.
[4] H.Balakrishan, V. N. Padmanabhan and R. Katz,”A
comparison of mechanisms for improving TCP
performance over wireless links”, in proceedings of ACM
SIGCOMM’96, palo alto, CA, Aug 1996,pp 256-269.
[5] Sarama Vangala, Miguel A. Labrodor,” Performance of
TCP over wireless Network with snoop protocol”
[6] Bikram Bakshi and N.H. Vaidya,”Improving performance
of TCP/IP over wireless networks”.
[7] R. Ramani, A. Karandikar, “Explicit congestion
notification in TC over wireless networks”.
[8] G. Holland and N. H. Vaidya, “Analysis of TCP
Performance over Mobile Ad Hoc Networks,”.
[9] J. P. Monks, P. Sinha and V. Bharghavan, "Limitations of
TCP-ELFN for Ad Hoc Networks,".
[10]D. B. Johnson, D A. Maltz, Y. Hu, “The dynamic souce
routing protocol for mobile ad hoc networks,”.
[11]Ajay bakre & B. R. Badrinath, “I-TCP: indirect TCP for
mobile hosts”.
[12]Raj Yavatkar & Namrata Bhagvat,”Improving end-to-end
performance of TCP over mobile envioronment”.
[13]Tom goff, James Moronski, D. S. Phatak, Vipul Gupta,
“Freeze-TCP: A true end to end TCP enhancement
mechanisms for mobile environment”
[14]P. Sinha,N. Venkitaraman,R Sivakumaran & V.
Bharghavan,”WTCP: realible transport protocol for wide
area network”.
[15]C. Parsa & J.J. Garcia-Luna-Aceves, “Improving TCP
congestion control over internets with heterogeneous
transmission media”.
[16]K. Brown & S. Singh, “M-TCP: TCP for mobile cellular
networks”.
[17]J. Liu, S. Singh, “ATCP: TCP for Mobile Ad Hoc
Networks”.
[18]Severio Moscola, Claudio casetti, Mario
Gerla,M.Y.Sanadidi & Ren Wang, “TCP
WESTWOOD:Bandwidth estimation for enhanced
transport over wireless links”.
[19]K. Chandran, R. Prakash, “Feedback based scheme for
improving TCP performance in Ad-hoc networks”.
[20]Dongkyun Kim, C-K Toh & Yanghee Choi “TCP-Bus-
Improving TCP performance in wireless Ad-Hoc
Networks”.
[21]Ranjeet V. Bidwe, “Different Issues and survey of
proposed solutions in TCP over Wireless Environment.”,
International Journal of Future Computer and
Communication, 2012.
[22] Ranjeet V. Bidwe, Amar R. Buchade, “TCP-P: A new
approach in wireless environment to solve issues
currently TCP facing” Second International Conference
on Computational Intelligence and information
Technology – CIIT 2012, Elsevier, 2012.
[23] Ranjeet V. Bidwe, Amar R. Buchade, “Improving
Performance of TCP in Wireless Environment using TCP-
P”, International Journal on Communication, Vol. 4, No.
1, 2013.
[24] Shamimul Qamar, Kumar Manoj, “Impact of Random
Loss on TCP Performance in Mobile Ad-hoc Networks
(IEEE 802.11): A Simulation-Based Analysis”, Arxiv
preprint arXiv9:1002.2403,arxiv.org, 2010.

TCP for Wireless Environments

More Related Content

What's hot (20)

Viewers also liked (18)

Similar to TCP for Wireless Environments (20)

Recently uploaded (20)

TCP for Wireless Environments