International Journal of Computer Science Research & Technology (IJCSRT)
ISSN: 2321-8827
QoS Optimization and Security Enhancements for VoIP in WLANs-Issues
Amel Chowdhury
Mathematics & Natural Sciences Department, BRAC University, Bangladesh
Abstract
We have addressed issues in implementing VoIP
services in packet switching networks, challenges to
enhance quality of service (QoS) and put several
solutions to improve VoIP performance in WLANs. To
provide competent QoS in VoIP system, several well
designed call admission control (CAC) mechanism
have been designed addressing issues such as
throughput, quality of voice, transmission delays etc.
But in practice, existing VoIP systems have not been
able to adequately apply and support these CAC
T they cannot provide QoS guarantees to VoIP
mechanisms which has been brought into one of the
SR
focuses of this paper. In the latter part of the article,
we present a brief survey of VoIP security academic
research providing a roadmap for researchers to find
IJC
out gaps among the existing capabilities of VoIP and
document the challenges which such infrastructure is
facing and analyze some of the solutions.
1.Introduction
Voice over IP (VoIP) is a critical real time
internet application which delivers voice packets over
the internet which reduces communication costs
immensely relative to telephone calls through public
switch telephone network (PSTN). Real time
transmission of voice packets has to follow very strict
requirements on delays and thus delay is an important
factor to impact the call quality. According to
International Telecommunication Union (ITU) a one
way delay maximum of 200 ms is acceptable in
Recommendation G.114. [1].
Call admission technique in traditional telephone
network follows that if sufficient link capacity is not
available, new calls are not admitted while current
calls remain unaffected. In IP networks, best effort
services are provided and regardless of the link
capacity, new calls are accepted whenever requested,
as a result channel congestion occurs along with
packet drops and delays. Hence, the necessity to
IJCSRTV1IS050073 www.ijcsrt.org
Vol. 1 Issue 5, October - 2013
design CACs. Other than delay requirements, VoIP
has differences over PSTNs such as voice compression
techniques are applied in VoIP networks which
increase bandwidth efficiency in a sense that the
remaining bandwidth is shared between other web
based traffic such as media and data application like
video, file share etc.
However, current situation with VoIP systems is that,
networks. The main reason is that none of the systems
are able to adequately support and implement the
designed CAC mechanisms. [2] Other challenges are
explained in [3] as: VoIP systems deployed in IEEE
802.11 WLANs use contention based medium access
control (MAC) protocol, the distributed coordination
function (DCF) which supports best effort traffic but
introduces large delays and delay jitter arbitrarily.
Thus it becomes unsuitable for real time applications
such as VoIP to provide strict QoS requirements.
Besides, PSTN and cellular networks have channels
dedicated to voice traffic, whereas in voice over
wireless LANs (VoWLANs) voice traffic is
multiplexed with data traffic and sent over WLANs.
This makes voice traffic unprotected. So mechanisms
to secure data involved in real time communication are
also a major research focus. Again when best effort
traffic load increases, the QoS for VoWLAN can be
severely degraded due to interference between each
other and due to reduction in system capacity. So
CACs have to be designed so that there is an optimum
condition between the high level of QoS in
VoWLANs and the high throughput of other traffic.
The basic building blocks installed at sender and
receiver ends required to run a PC-PC VoIP services
are given in Figure 1.The speech signal is encoded at
the speech encoder which includes acquisition,
sampling and compression operations. The analog
voice signals are sampled at a fixed frequency where
62

each sample is encoded. The bitstream produced at the
encoder is placed in packets following specific filling
procedures to ensure the compatibility between
heterogeneous devices. Each speech encoder defines a
packetization mechanism which is managing a trade-
off between end-to-end delays and packet rate. At the
receiver side, a playback algorithm determines
determines the playback time of each received speech
frames. [4] Other than voice channels, there is a
signaling channel which performs caller and callee
identification, call redirection, edge device
configuration and QoS negotiation. [5]
T
FIGURE 1. Basic building blocks of a VoIP
SR
system
To discuss the QoS mechanisms, we do not present
IJC
any original technical research or summarize all kinds
of mechanisms developed so far. Rather we focus on
some major QoS issues which deal with capacity
increase, delay reduction, throughput and issues
related to feasibility to provide such guarantee after
integration to any existing VoIP system. Any other
QoS mechanisms which have been implemented or is
under study, we try to put as much as possible in the
„related work‟ section. On the latter part of the paper,
we bring on the other essential part of a VoIP system
which are the security concerns. We discuss some
common VoIP system or computer attacks and then
put a brief survey on solutions to prevent such attacks
covering the major issues.
Rest of the article organization is as follows. In section
2, we discuss related works for QoS implementation in
VoIP in IEEE 802.11 LANs. In section 3, we present
three mechanisms selectively to deal with different
QoS factors in detail and discuss their advantages over
other design mechanisms. In section 4, the general
security concerns are given followed by some related
work to mitigate some specific attack in section 5. In
section 6 we draw conclusion to this paper.
IJCSRTV1IS050073 www.ijcsrt.org
International Journal of Computer Science Research & Technology (IJCSRT)
ISSN: 2321-8827
Vol. 1 Issue 5, October - 2013
2. Related Work
In the recent past, several research works have
addressed the performance and QoS issues of
supporting VoIP over internet. In [6], to reduce
congestion and overall end to end delay, switching
among multiple paths have been proposed and
methods to recover from packet loss in VoIPs in
overlay networks in [7]. VoIP systems over IEEE
802.11 WLANs have been studied in [8][9] where the
delay and loss characteristics under PCF and DCF
mode. In [10], an analytical studies on the number of
calls that can be supported in a single hop WLAN is
presented. Other performance schemes have been
proposed to improve the VoIP quality over WLAN in
[11] which describes the use of dedicated queue to
provide higher priority to VoIP traffic over data traffic
all the time.
Another common scheme to improve end to end
throughput for various applications on multihop
networks is packet aggregation or concatenation. In
[12], packet concatenation has been proposed to
reduce 802.11 MAC overhead between a single source
destination pair. There are other packet aggregation
schemes described for adhoc networks where in some
cases they introduce additional delays at each
intermediate nodes and in some other adaptive
algorithms as in [13] describes that when enough
packets are not available to be concatenated, the
bandwidth usage becomes inefficient.
3.QoS Mechanisms For VoIP
3.1 VoIP Multiplex-Multicast Scheme
In [14], a voice multiplex-multicast (M-M) scheme
has been proposed to overcome the huge overhead
produced in VoIP applications over WLANs. The
main idea of packet M-M scheme is to concatenate
several packets from different downlink stream to one
larger packet.
As shown in Figure.2, the downlink VoIP packets at
first goes through a multiplexer (MUX) which
compresses RTP, UDP and IP header of each packet
into a miniheader which combines multiple packets
into a single packet. This multiplexed packet is then
transmitted across the WLAN through access points (AP)
with a multicast IP address. At the receiver end, the
DEMUX retrieves the VoIP payload and extracts its
original RTP, UDP and IP header and necessary
destination information. . The original recipient of a
63

(a)
(b)
Figure 2. a) Traffic flows in an ordinary VoIP
scheme
b) Traffic flows in the VoIP M-M scheme
T utilization at the optimal point. The CARC mechanism
SR
certain VoIP packet identifies it using an ID which is
added in the miniheader in prior. Finally, the DEMUX
assembles all the data of the VoIP packet into its
IJC
original form and forwards it to the application. From
Figure.2 we can say, in this method the number of
VoIP streams can be reduced from 2n to n+1, where n
is the number of VoIP sessions. The header
compression for RTP, UDP and IP is very agreeable
with such systems as most of the fields of RTP, UDP
and IP headers do not change throughout the entire
lifetime of an RTP stream. With this scheme these
headers are replaced by as most of the fields of a 2
Byte miniheader, the compression mechanism is given
detailed in [15].
The M-M scheme claims to achieve 80%-90% higher
capacity than an ordinary VoIP over WLAN. This is
explained in [14]. When M-M scheme is applied, the
RTP, UDP and IP headers are compressed to 2 Byte,
and the multiplexed packet multicasted, so there is no
overhead for the receiver. So, the total number of
supported sessions increases.
The delay produced by this scheme is much less than
125 ms whereas ITU recommended acceptable range
for a one way delay is (0-150)ms. The M-M scheme
addresses interference problem of VoIP traffic with
the TCP traffic by using priority queuing (PQ) for the
VoIP traffic. But with PQ, packet loss rate remains
IJCSRTV1IS050073 www.ijcsrt.org
International Journal of Computer Science Research & Technology (IJCSRT)
ISSN: 2321-8827
Vol. 1 Issue 5, October - 2013
very high. The problem is dealt by introducing MAC
layer multicast priority scheme (MMP). With MMP,
when an AP has a multicast frame, it waits for a
multicast interframe space (MIFS) period before
transmission rather than the DIFS period and CW-
backoff. This reduces the packet loss due to collisions.
The MIFS is set to a value less than the DIFS but
larger than the SIFS.
There have been other schemes as [16-18] which
almost have 100% reliability by using different
transmission strategies but those are not scalable and
causes unacceptable delays in VoIP.
3.2. Call Admission and Rate Control
(CARC)scheme:
The call admission and rate control (CARC) scheme
assumes channel busyness ratio 𝑅𝑏 as the control
metric which is the ratio of the time the channel
remains busy during both successful transmission and
collision time. Here, 𝐵𝑢 is assumed to be the channel
tries to keep 𝑅𝑏 close to 𝐵𝑢 to ensure a good QoS level
along with high throughput. The CARC admits a new
voice calls only if there is sufficient bandwidth
available. An upperbound 𝐵𝑀 for bandwidth
reservation for voice traffic is assumed which is set to
80% of channel utilization 𝐵𝑢 of the WLAN. The rest
20% is kept to treat the best effort TCP traffic. The
bandwidth requirement for a voice call is converted to
the rate of the voice traffic. For example, if „r‟ is the
average rate of the voice traffic, 𝑟𝑝𝑒𝑎𝑘 is the peak rate
and „l‟is the length of the packet in terms of bits, then
channel utilization and its peak value ‘u’ is obtained
by [3]-
𝑟 𝑟 𝑝𝑒𝑎𝑘
𝑢 = × 𝑇𝑠𝑢𝑐 and 𝑢 = × 𝑇𝑠𝑢𝑐
𝑙 𝑙
where 𝑇𝑠𝑢𝑐 = time of successful transmission of a
packet including all the interframe spacing and ACK
messages. The total bandwidth occupied by all
admitted flows at the AP of a WLAN is given by
aggregated 𝑢 and 𝑢𝑝𝑒𝑎𝑘 denoted by 𝑢𝐴 and 𝑢𝑝𝑒𝑎𝑘𝐴
when any node receive a voice call request, it converts
the bandwidth requirement in the form of (𝑢, 𝑢𝑝𝑒𝑎𝑘 )
and compares if 𝑢𝐴 + 𝑢 ≤ 𝐵𝑀 and 𝑢𝑝𝑒𝑎𝑘𝐴 + 𝑢𝑝𝑒𝑎𝑘 ≤
𝐵𝑢 . If it is satisfied then AP issues a connection
admitted message, otherwise a connection rejected
message is generated.
64
 International Journal of Computer Science Research & Technology (IJCSRT)
ISSN: 2321-8827
Vol. 1 Issue 5, October - 2013

The rate control regulation controls the packet bandwidth usage. A call is admitted whenever the used
sending ratio by using another parameter called „s‟ bandwidth resource and the requested bandwidth
(0<s<1) which is the allowed share for best effort resource do not exceed this limit which is prefixed
traffic for a node to contend for the shared channel. offline at the time of configuration. The call admission
There traffic shall use the residual bandwidth left by procedure is executed basically upon two modules
the real time traffic. Allowed share „s‟ is defined by which are utilization computation module and
𝑡 admission decision making module. The utilization
𝑇 = 𝑝 where 𝑡𝑝 is the time a successful transmission
𝑠 computation module performs the delay analysis and
of packet „p‟ will last over a channel and T is the time
computes bandwidth utilization. The utilization
between two consecutive packets that passes to the
module have two kinds of submodules which are link
MAC layer.[3] According to CARC scheme, the
utilization based call admission control (LU-CAC) and
parameter „s‟ adjusts its value dynamically according
site utilization based call admission control (SU-
to the network condition. The adaptation procedure
CAC). The computed utilization is placed in either of
goes as follows:
LU-CAC or SU-CAC, then the admission decision
making module decides for each incoming call on the
-if the channel busyness ratio 𝑅𝑏 < 𝐵𝑀 , the channel is
basis of the received data.
assumed to be underloaded, in that case the rate
control mechanism adopts a multiplicative-increase The main task of LU computation submodule is to
law, multiplying „s‟ by the ratio of 𝐵𝑢 to 𝑅𝑏 : compute maximum link utilization for LU-CAC by a
𝐵
𝑠←𝑠× 𝑢 utilization verification procedure which is given in
𝑅𝑏
Figure 3.[14] The delay analysis technique determines
This way 𝑅𝑏 quickly converges to 𝐵𝑢 . When worst case delays of deadline violation probabilities
𝐵𝑀 ≤ 𝑅𝑏 < 𝐵𝑢 , the channel is considered as assuming a worst case combination of flows and it is
T performed by LU-CAC both in deterministic and
moderately loaded. Then the rate control mechanism
statistical manner.
SR
𝑡
adopts an increase law [3]: 𝑠 ← 𝑠 + 𝑝 𝛿 where 𝛿 is the
𝑠
𝑡
increase factor and 𝑝 is the interval between two
𝑠
consecutive packets passed to the MAC layer. Here
IJC

the increase in share occurs in proportional to the

interval between two consecutive packet transmission
and is assumed to have a fair allocation of channel
share.

When 𝑅𝑏 > 𝐵𝑢 , the „s‟ factor has to be decreased,
according to [3] the multiplicative decrease law is
𝐵
given by- 𝑠 ← 𝑠 × 𝛾 × 𝑢 where 𝛾 is the decrease
𝑅𝑏
factor and 0 < 𝛾 ≤ 1. In these equations, the factors 𝛿
and 𝛾 control the convergence speed. This way the
bandwidth is attempted to be utilized in an optimized
manner and collision occurance is handled well
enough too. To alleviate collisions as much as
possible, the CARC adopts a packet defer procedure
separately. From the simulation results in [3], it shows
the CARC method allows delays for voice traffic at
around 70-80 ms maintaining a very high throughput
and avoiding collisions.Thus it provides good
statistical guarantee for QoS and also does not need
upgradation on the firmware of MAC controller chip.
3.3. Utilization based call admission control
This is a kind of call admission control mechanism
which uses predefined utilization limit in terms of
Figure 3. Utilization verification process
3.3.1. Utilization based Deterministic Delay
Analysis. Assuming the potential end to end delay for
a certain network topology and the bandwidth
utilization is known, the worst case queuing delay 𝑑𝑘
suffered by any voice packet with highest priority at
the buffer of o/p link „k‟ is bounded by-
( 𝐶)
𝑐 −1 𝜍
𝑑𝑘 ≤ 𝑘 𝑢𝑘 ( + 𝑌𝑘 ), where 𝑐𝑘 = 𝑗𝜖𝑙 𝑘 𝑗 𝐶 ,
𝑐 𝑘 −𝑢 𝑘 𝜌 𝑘
𝑌𝑘 = 𝑚𝑎𝑥𝑅𝜖 𝑠𝑘 𝑠𝜖𝑅 𝑑𝑠 , 𝐿𝑘 is the set of all the i/p links
of o/p link „k‟, and 𝑆𝑘 is the set of all subroutes used
by voice packets with highest priority upstream from
o/p link ‘k’. [19]
3.3.2. Utilization based Statistical Delay Analysis.
When deadline requirement is probabilistic, we can
find delay probabilities as-

IJCSRTV1IS050073 www.ijcsrt.org 65
 International Journal of Computer Science Research & Technology (IJCSRT)
ISSN: 2321-8827
Vol. 1 Issue 5, October - 2013

If 𝑑𝑘 is a random variable and 𝐷𝑘 is denoted as its In this scheme the major advantage which has been
deadline, the violation probability of delay for any demonstrated over any other QoS design mechanisms
voice packet with the highest priority suffered at the is that this have been integrated experimentally with
buffer of o/p link „k‟ is bounded by- existing VoIP system such as Cisco VoIP system and a
satisfactory QoS level has been observed. Whereas,
𝑃 𝑑𝑘 > 𝐷𝑘 ≤ currently Cisco systems used to perform resource pre
1 1 − 𝑢𝑘 𝐷𝑘 𝐷𝑘 allocation in an adhoc manner, hence no QoS could be
exp −24 2 𝜍 , 𝑢𝑘 ≥ 𝜍 guaranteed. [20]
2𝜋 𝑢𝑘 𝜌 𝜌
2
1 1 − 𝑢𝑘 𝐷𝑘 𝐷𝑘
exp −6 3 𝑢𝑘 + 𝜍 , 𝑢𝑘 < 𝜍 4.Security Issues in VoIP
2𝜋 𝑢𝑘 𝜌 𝜌 Security issues are categorized in three categories
confidentiality, integrity and availability, in other
The end to end deadline violation probability can be words current VoIP systems has to put a compromise
bounded by-[2] on these factors. Confidentiality threats include
exposing the contents of conversation between two
𝑃 𝑑 𝑒2𝑒 > 𝑘𝜖𝑅 𝐷𝑘 ≤ 1 − 𝑘𝜖𝑅 (1 − 𝑃{𝑑𝑘 > 𝐷𝑘 }) parties, integrity threats indicate the ability to trust the
identity of a caller, of a message or the identity of the
which depend on the link utilization 𝑢𝑘 , the parameter
recipient or the call record logs. [41] Availability
for voice traffic like burst size 𝜍 and average rate threats impact the ability to initiate a session.
𝜌.The main task of site utilization computation
submodule is to optimize the overall bandwidth Attacks such as the denial of service (DoS) attempted
utilization to sites defined in [2] as- by an attacker will prevent the VoIP system from its
Maximize 𝑅 𝑢𝑅 (overall bandwidth)
T normal operating condition and no user will be able to
receive or make a call. Eavesdropping is leakage of
SR
Subject to 𝑅𝜖𝑘 𝑢𝑅 ≤ 𝑢𝑘 (bandwidth preallocation someone‟s conversations by getting monitored by the
for each pair of site constrained by bandwidth attacker secretly. In this process the attacker can
limitations). collect data from both the parties involved in the call,
IJC

which he can use it later by replaying the conversation

𝑢𝑅0 ≤ 𝑢𝑅 ≤ 𝑢1𝑅 for each route ‘R’(user requirement for
bandwidth preallocation for each pair of sites).
Here 𝑢𝑘 is the maximum bandwidth of link „k‟
allocated to voice traffic, 𝑅𝜖𝑘 represents all routes
among any pair of sites R is going through link , 𝑢𝑅 is
the bandwidth (BW) for R allocated to voice traffic
and 𝑢𝑅0 & 𝑢1𝑅 are the lower and upper bandwidth
bounds for R . SU-CAC does not adopt BW allocation
dynamically according to the network condition as it is
pre allocated. It can be done through LU-CAC, but its
implementation can be complex. In SU-CAC, sites are
assumed to be any location to the Call Manager or any
zone to the Gatekeeper. After performing the
maximum BW utilization computation, it is fed into
the admission decision making module which
implements call admission control (CAC)
mechanisms. These modules are integrated with
existing VoIP systems through and integration
components (IC).[2] In this work, they discuss another
component called QoS Manager (QoSM) which
provides interface to control and monitor components
and cooperates with peer QoSMs. It also provides
registration and coordination between distributed
agents within a QoS domain.
and use the contents for any illicit purpose.
Confidentiality and integrity of any session can be
severely hampered by the attacker as they are able to
not only listen to the conversation but also alter it and
make the receiver hear a message different from what
the sender is sending. This situation is also called the
man in the middle attack. Other than these, attackers
can redirect the victims call to any other attacker‟s PC
or phone. An attacker can also change the caller ID of
the victims phone. There are other security issues such
as toll fraud which makes the victim pay for calls
which they did not make, in other words attackers can
place unwanted amount of calls to any phone number.
To fight back all these threats several security
protocols have been designed and implemented so far.
Such as IP security (IPSec) provides mechanisms for
authentication and encryption. Within IPSec other
security mechanisms are implemented such as
certification authority (CA) and DNS secure
(DNSSec) protocol. As VoIP phones need basic
configuration information to get into VoIP system, at
the time of manufacture, the IP phones are
preconfigured with public key of various configuration

IJCSRTV1IS050073 www.ijcsrt.org 66

servers which is a common authentication mechanism
for the configuration server. [41] An extension to the
RTP protocol called secure real time protocol (SRTP)
provides authentication and confidentiality services for
the message carried by the RTP protocol. This adds a
lower overhead comparing to the packet size and also
minimizes the number of key pairs shared by two
communicating nodes.[42] Besides the
implementation of security protocols there are many
related works developed to defend specific VoIP threat
issues which we discuss below.
5. Defense Mechanisms of VoIP Attack
5.1.Sybil Attack
[21, 29] suggest to reduce the effects of Sybil attacks
by applying mechanisms to reduce the amount of
information received by malicious nodes. Sybil effects
are known as addition of fake identities to nodes
which are supposed to join the overlay with their
regular numeric identifiers (IDs) to indicate their
positions, what information they store and whom they
serve. Normally, the contents and services are
T
replicated across multiple nodes to provide a degree of
robustness.[22] Attackers use this flexibility and
SR
attempt to destroy the actual purpose of replication.
Several ideas [23,24,25] tell to vary the position of the
nodes used for routing to defend these attacks and to
IJC
avoid a trust bottleneck. [26] In [27] they suggest to
place the nodes with diametrically opposite IDs in the
distributed has tables (DHT) ID space. With
diametrically opposite IDs the nodes will be placed
furthest away and attackers will need more resources
to attack in two positions far apart.
5.2. Loss of Data Integrity
This is described as insertion of wrong information or
corrupted chunk of datas in file sharing application.
[27] For file sharing, these corrupted datas will be the
files themselves which will be publicly available and
each user usually looks for thousands of files stored in
the DHTs. In real time communication, users register
only a limited number of locations at a time and the
overlays use only a portion of the available resources.
Attackers need very little to pollute from these limited
locations resulting in significant reliability problem. In
conventional peer-to-peer overlays, protocols as Bit-
Torrent uses moderators to remove bogus files and
uses SHA-1 algorithm to verify the integrity of hash of
each piece of a file.[27, 28]
IJCSRTV1IS050073 www.ijcsrt.org
International Journal of Computer Science Research & Technology (IJCSRT)
ISSN: 2321-8827
Vol. 1 Issue 5, October - 2013
5.3. Loss of Routing Integrity
Providing wrong responses to queries and routing
through malicious nodes is a similar significant
problem as insertion of polluted data and is known as
loss of routing integrity. It is important to ensure that
the correct location that a user requests for, is being
returned. In [29], the idea is that a user should choose
asymmetric key pair and hash its public key to
generate the URI and then signs its own location with
a private key and registers for (URI, IP, signature,
public key). Any other node enquiring for this location
will receive this quadruplet and will check the
integrity matching it with the public key.
5.4. Resource Drained DoS Attacks
In [30] a novel way to detect resource drained attack
has been presented. In a SIP session several basic
messages to initiate, forward, acknowledge and then
terminate (INVITE, 200 OK, ACK, BYE) sessions are
used between user agent client (UAC), user agent
sender (UAS) and SIP proxy servers. There is a
session timer as SIP extension header which when
expires, the resource retained by the proxy server for a
session is released. It is possible for an attacker to
manipulate the session timer and act both as UAC and
UAS and send SIP INVITES to initiate an attack. The
proxy servers hold resources according to the session
timer, let the attacker disconnect from the network and
if timer is longer, the attacker can initiate another
session after some time and make further reservation
of resources on the proxy servers. Thus the capacity of
the SIP proxy servers to process normal messages are
hampered severely leading to denial of service. In
[30], performing normality tests, the session timer
sequence has been observed, when under attack the
timer values can hardly be characterized, so they
determine a threshold statistically. If the timer
statistics exceed the threshold then the session is
rejected assuming an attack has been detected. A DoS
attack can occur against a particular node by
bombardment of huge amount of queries. To mitigate
this [29] gives a solution where peer nodes vary the
target nodes used in queries.
5.5. Man in the Middle Attack
When malicious nodes can return the IDs of other
malicious nodes when queried of a particular ID is
known as man in the middle attack. When the
requester establishes a session with the malicious
67

node, it gives a poisoned reply and this can go on and
on.[26] A not so effective approach to solve this is
presented in [32] which says to employ iterative
routing and to check the ID of every routing hop and
finally is expected to reach the desired node.
5.6. Flooding Attack
In [32], it has been explained that the SIP protocol lets
an incoming request to branch to multiple outgoing
requests each for different UASs. Less than ten
messages can generate 271 messages occurring
massive flooding attack with valid SIP requests. [33]
The SIP routing occurs from proxy to proxy servers
based on the routing headers, but this process
possesses vulnerabilities like-[34]
-attackers can manipulate routing headers
-proxy servers route without call-route or global route
knowledge.
-the HTTP digest based authentication to protect SIP
messages is not an end-to-end security model as
intermediate proxies change certain fields
-in this kind of SIP authentication mechanism, few SIP
fields are protected leaving most of them unprotected
T 7. References
along with the messages while they are routing
SR
through proxies.
In [32], a language dedicated to attack recognition has
been presented called VeTo. VeTo has three features
IJC
combining context, definition and prevention. The
context blocks exhibit the vulnerabilities surrounding
environment properties. The definition block provides
the vulnerabilities related to the behavior such as SIP
messages and its respective fields. The prevention
block describes vulnerable behavior from the context
and generates a response action. The full
implementation issues and the language specifications
are given in [35, 36].
5.7. Loss of Confidentiality
Unlike publicly shared files, P2P RTP sessions
requires absolute privacy. Following SRTP [37] voice
messages are encrypted and transmitted through
channels. In [38], they give the example of
stenographic insertion of a second voice channel in a
SIP based VoIP conversation which has the potential
to leak out an encrypted conversation to the secondary
channel or hide the true communication content from
an eavesdropper. They present some possible detection
and mitigation methods. In [39] they apply machine
learning techniques to determine language spoken in a
VoIP. For prevention, they suggest to use block
ciphers for content encryption. In [40] they use Hidden
Markov Models to identify specific phrases in the
IJCSRTV1IS050073 www.ijcsrt.org
International Journal of Computer Science Research & Technology (IJCSRT)
ISSN: 2321-8827
Vol. 1 Issue 5, October - 2013
encrypted voice stream with a 50% average accuracy
and 90% to certain phrases.
6.Conclusion
In this article we basically put a brief survey on two
factors essential to a VoIP systems- ensuring quality
of service and ensuring security or confidentiality
throughout the end to end communication system.
Implementation of QoS is moving towards a more and
more satisfactory level day by day but VoIP
deployment still faces great challenges regarding
malicious attacks and requires numerous counter
measures which has to be a continuous process for
future implementations. Although there are standards
for VoIP protocols and services, security management
of VoIP systems require continued evolution of these
services and protocols to tighten security. We
conclude the paper hoping the survey can ease the task
of conducting further research in VoIP security and
QoS enhancement.
[1]“One-Way Transmission Time (Recommendation
G.114),” Int’l Telecomm. Union (ITU), 1996.
[2] S. Wang, Z. Mai, D. Xuan and W. Zhao, “Design and
Implementation of QoS-Provisioning System for Voice over
IP,” IEEE Trans .Parallel & Dist. Sys.,vol.17, no.3, pp.
276-288, Mar. 2006.
[3] X. Chen, H. Zhai, and Y. Fang, “Providing Statistical
QoS Guarantee For Voice Over IP In The IEEE 802.11
WirelessLANs,” IEEE Wireless Commun., pp. 36-43, Feb.
2006.
[4] S. Jelassi, G. Rubino, H. Melvin, H. Youssef and G.
Pujolle,“Quality of Experience of VoIP Service:A Survey of
Assessment Approaches and Open Issues,” IEEE
Commun.Surveys &Tutorials vol. 14, no. 2, pp. 491-512,
2012.
[5] H. Schulzrinne and E. Wedland, “Application-layer
mobility using SIP,” SIGMOBILE Mob. Comput. Commun.
Rev., vol. 4, pp 47-47, July 2000.
[6] S. Tao, K. Xu, A. Estepa, T. Fei, L. Gao, R. Guerin, J.
Kurose and D. Towsley, “Improving VoIP Quality
Through,” In Proc.of Infocom 2004, Mar. 2004.
[7]Y. Amir, C. Danilov, S. Goose, D. Hedqvist and A.
Terzis,“1-800-overlays: Using overlay networks to improve
VoIP Quality,” In Proc. of NOSSDAV, 2005, 2005.M.
[8]Veeraraghavan, N. Cocker, and T. Moors, “Support of
voice services in IEEE 802.11 wireless LANs, ” In Proc. of
INFOCOMM 2001.
[9]D. Hole and F. Tobagi, “Capacity of an IEEE 802.11b
wireless LAN supporting VoIP, ” In Proc. of ICC 2004.
[10]S. Garg and M. Kappes, “Can I add a VoIP call?,” IEEE
ICC, (Anchorage, Alaska), 2003.
68

[11]J. Yu, S. Choi and J. Lee, “Enhancement of VoIP over
IEEE 802.11 WLAN via dual queue strategy, ” In Proc.of
ICC 2004.
[12]Y. Xiao, “Concatenation and Piggyback Mechanisms for
the IEEE 802.11 MAC,” IEEE WCNC, 2004
[13] A.Jain, M. Gruteser, M. Neufeld and D. Grunwald,
”Benefits of packet aggregation in ad-hoc wireless
network,”Tech. Rep.CU-CS-960-03, Department of
Computer Science, University of Colorado at Boulder, 2003.
[14] W. Wang, S. C. Liew and O. K. Victor, “Solutions to
Performance Problems in VoIP Over a 802.11
WirelessLAN, ”IEEE Trans. Vehicular Tech.,vol.54,no.1,
pp366-384, Jan.2005.
[15] H. P. Sze, S. C. Liew, J. Y. B. Lee and D. C. S. Yip,
“A multiplexing scheme for H. 323 voice over
IPapplication,”IEEE J. Sel. Areas Commun, vol. 20, no.
9,pp. 1360-1368, Sep. 2002.
[16] J. Kuri and S. K. Kasera, ” Reliable multicast in multi
Access wireless LANs,” Proc.INFOCOMM ’99, vol. 2, pp.
760-767, Mar. 1999.
[17] M.T. Sun, L. Huang, A. Arora and T. H. Lai, “Reliable
MAC layer multicast in IEEE 802.11 wireless
networks,”Proc.Int. Conf. Parallel Processing, Aug.2002,
pp.527-536.
[18] K. Tang and M. Gerla, “MAC layer broadcast support
in 802.11 wireless networks,”Proc.MILCOMM’00, vol.1,pp
T
.544-548, Oct. 2000.
SR
[19] S. Wang, D. Xuan, R. Bettati and W. Zhao, “Providing
Absolute Differentiated Services for Real-Time Applications
in Static-Priority Scheduling Networks,”IEEE/ACM Trans.
Networking, vol. 12, no. 2, pp. 326-339, 2004.
IJC
[20] J. Davidson et al.,Deploying Cisco Voice over IP
Solutions. Cisco Press, 2002.
[21] R. Baumann, S. Cavin and Schmid, “Voice Over IP-
Security & SPIT,” KryptDet Report FU Br 41,Swiss Army,
Aug/Sep.2006.
[22] Angelos D. Keromytis, “A Comprehensive Survey of
Voice over IP Security Research,” IEEE Commun.Surveys
Tutorials,vol. 14, no. 2, 2012.
[23] I. Stoica, R. Morris, D. Karger, M. F. Kaashoek and H.
Balakrishnan,”Chord: A scalable peer-to-peer lookup service
for internet applications,” SIGCOMM, 2001.
[24] P. Maymounkov and D. Mazi, “Kademlia: A peer-to-
peer Information system based on the xor metric,” First
International Workshop on Peer-to-Peer Systems, Mar.
2002.
[25] A. Rowstron and P. Druschel, “Pastry: Scalable,
distributed object location and routing for large scale peer-
to-peer systems,” 18th IFIP/ACM International Conference
on Distribute Systems Platforms (Middleware 2001), Nov
2001.
[26] G. Danezis, C. Lesniewski-Laas, M. F. Kaashoek and
R. Anderson, “Sybil-resistant DHT routing,” Tenth
European Symposium on Research in Computer Security,
vol.3679,Sep. 2005.
[27] D. Chopra, H. Schulzrinne, E. Marocco and E. Ivov,
“Peer-to-Peer Overlays for Real-Time Communication:
Security Issues and Solutions,” IEEE Surveys Tutorials, vol.
11, no. 1, pp.4-12, 2009.
IJCSRTV1IS050073 www.ijcsrt.org
International Journal of Computer Science Research & Technology (IJCSRT)
ISSN: 2321-8827
Vol. 1 Issue 5, October - 2013
[28] X. Zhang, S. Chen and R. Sandhu, “Enhancing data
aunth- enticity and integrity in P2P systems,”Internet
Computing,Sep. 2005.
[29] J. Seedorf,”Using cryptographically generated SIP-
URIs to protect the integrity of content in P2P-SIP, ” VoIP
Security Workshop, Jun. 2006.
[30]J. Tang,Y Hao,Y.Cheng and C.Zhao,“Detection of
Resource-
Drained Attacks on SIP Based Wireless VoIP NetWorks,” in
Proc. IEEE GLobecom 2010.
[31] W.-K. Poon and R. K. C. Chang, “Robust forwarding in
structured peer-to-peer overlay networks,” SIGCOMM,
Aug. 2004.
[32] ] A, Lahmadi and O. Festor, “VeTo: An Exploit Prevent
-tion Language from Known Vulnerabilities in SIP Services,
”IEEE/IFP Network Operations and Management
Symposium-NOMS 2010. pp.216-223.
[33] R. Sparks, S. Lawrence, , A. Hawrylyshen and B
Campen,“Addressing an Amplification Vulnerability in
Session initiation Protocol (SIP) Forking Proxies,”
RFC5393(Proposed Standard),Dec 2008.
[34]D. Schwartz and J. Barkan, “End-to-end route
management in session initiation protocol,
http://tools.ietf.org/html/draft-schwartz-sip-routing-
management-00, Feb. 2006.
[35]A. Lahmadi and O. Festor, “Secsip: A stateful firewall
for sip-based networks,” In the Proc.of 11th IFIP/IEEE
International Symposium on Integrated Network
Management,IM09, Jun. 2009.
[36]A. Lahmadi and O. Festor,”VeTo:Reference Manual.
Technical Report, ” Loria-INRIA Nancy Grand Est.
Research Center, Jul. 2009.
[37] M. Baugher, D. McGrew, M. Naslund, E. Carrara and K
Norman, “The secure real-time transport protocol (SRTP),”
RFC 3711 (Draft Standard), Mar. 2004. [Online] Available:
http://www.ietf.org/rfc/rfc3711.txt
[38] T. Takahashi and W. Lee, “An Assessment of VoIP
Covert Channel Threats,” in Proc.3rd International
Conference on Security and Privacy in Communications
Networks (Secure Comm,) pp. 371-380, Sep. 2007.
[39]C. V. Wright, L. Ballard, F. N. Monrose and G. M.
Masson, “Language Identification of Encrypted VoIP
Traffic: Alejandro y Roberto or Alice and Bob?,” in Proc.
16th USENIX Security Symposium, pp. 1-12, Aug. 2007.
[40] C. V. Wright, L. Ballard, F. N. Monrose and G. M.Mass
-on, “Spot Me If You Can: Recovering Spoken Phrases in
Encrypted VoIP Conversations,” in Proc.IEEE Symposium
on Security and Privacy, pp. 35-49. May 2008.
69