Password based Secure Authentication Methodology for
Wireless Sensor Network
Gul N. Khan and M. Zulfiker Ali
Department of Electrical and Computer Engineering, Ryerson University,
350 Victoria Street, Toronto, ON M5B2K3, Canada
Keywords: Authentication Protocol, Secure Authentication, Sensor Data Security, Wireless Sensor Network.
Abstract: Wireless Sensor Network (WSN) is a promising solution for next generation real-time monitoring
applications due to its ubiquitous nature, ease of deployment and wide range of applications. The main
requirements of a secure WSN architecture are confidentiality, integrity and authentication. User
Authentication for wireless sensor networks is a fundamental issue in designing secure WSN systems,
where legitimate users are allowed to login and access data from the sensor and gateway nodes. A large
number of dynamic strong password and two-factor authentication solutions have been proposed. However,
all of these protocols rely on network synchronization and suffer from replay and many logged in users with
same login ID threats. We present a secure authentication protocol that avoids replay and multiple login
attacks. Our scheme eliminates the need for network synchronization. The scheme is analysed and evaluated
for various attacks using SystemC and it provides enhanced security for no extra computing at sensor nodes.
1 INTRODUCTION
Wireless Sensor Networks (WSNs) are used to
collect and process environmental data such as
temperature, humidity, light, seismic activities,
images, etc. WSNs have attracted many researchers
due to their ubiquitous nature, easy deployment and
a wide range of applications. Wireless networks with
thousands of tiny sensors having low processing
capabilities and limited memory play an important
role for economical solutions of some challenging
problems including real-time traffic monitoring,
building safety, object tracking, wildlife monitoring,
etc. In general, most of the queries in WSN
applications are issued at the base stations or
gateway nodes of the network. However, one can
foresee that there are greater needs to access the
real-time data within a WSN. For some applications,
the collected data is valuable and confidential, while
many applications require integrity and
confidentiality of the collected data along with the
user privacy. Security measures are incorporated to
protect the data access and to prevent intruders from
gaining the data access. If the data is made available
to the user on demand then user authentication must
be ensured before allowing its access.
Many WSN related user authentication schemes
have been proposed that are suitable for wireless
mobile and low-power devices. IEEE standard
802.15.4 (2003) is the most appropriate
communication scheme for low power sensor
networks. Sastry and Wagner (2004) observed the
merits and limitations of security aspects related to
WSNs.
A number of researchers have focussed on the
user authentication schemes suited to WSNs.
Benenson et al., (2004) introduced an n-
authentication protocol in which the authentication
succeeds if the user can successfully authenticate
with a subset of n-sensors. Benenson et al., (2005)
proposed a public key based user authentication
protocol with elliptic curve cryptography. Wong et
al., (2006) proposed a strong password-based
dynamic user authentication scheme. Tseng et al.,
(2007) pointed out some weaknesses of Wong et
al.’s (2006) scheme and proposed an improved
dynamic user authentication method to enhance
security. Lee (2008) also analyzed this scheme and
proposed two simple dynamic user authentication
methods. Ko (2008) pointed out that Tseng’s et al.,
(2007) scheme suffers from replay and forgery
attacks. Ko’s user authentication scheme (2008) also
provides mutual authentication. However, Vaidya et
35
N. Khan G. and Ali M..
Password based Secure Authentication Methodology for Wireless Sensor Network.
DOI: 10.5220/0004307400350041
In Proceedings of the 3rd International Conference on Pervasive Embedded Computing and Communication Systems (PECCS-2013), pages 35-41
ISBN: 978-989-8565-43-3
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
al., (2009) argued that most of these schemes cannot
preserve user anonymity. Das (2009) introduced the
two-factor authentication concept to resist many
logged in users with the same login-ID and stolen-
verifier attacks. Khan and Alghathbar (2010)
incorporated enhanced security patch into Das’s
(2009) scheme. Vaidya et al., (2010) pointed out that
several security pitfalls still exist in these schemes
and proposed an improvement method that results in
a high level of robustness and security. Zhou et al.,
(2011) have proposed a new dynamic user
authentication method employing nonce in place of
timestamps.
Two-factor user authentication schemes are
expensive but password-based authentication can be
easily integrated. To the best of our knowledge
Vaidya et al.’s scheme (2009) is the latest variant of
strong password-based user authentication method.
However, the scheme allows data access from sensor
nodes without gateway’s notice that makes it
vulnerable to gateway bypass attacks due to node
compromise attack. Moreover, the scheme does not
provide mutual authentication between a user device
(UD) and the gateway node (GW). All password-
based schemes require strict time synchronization,
which increases the network overhead and makes
them vulnerable to replay attack within a time
interval. Their methods do not resist many logged in
users with same ID attacks. In this paper, we
propose a robust and secure user authentication
method that resists gateway bypass, replay and many
logged in user with the same ID attacks.
2 AUTHENTICATION SCHEME
The notations used in this paper are listed in Table 1.
Our scheme assumes a typical WSN setup shown in
Figure 1. Authorized users can access the WSN
anywhere in the network with mobile devices by
communicating with a sensor node. Before issuing
any queries, a user must register at the gateway
(GW) node. Upon successful registration, the user
can submit its query to a sensor node (SN) any time
within a predefined period. The allocation of a secret
key to the user and sensor nodes is dynamic that
takes place during registration. A flag is maintained
for each user in the GW to prevent multiple logins.
We assume that the registration process is carried
out in a secure manner and the database stored at the
GW node is secure. In our user authentication
scheme, we used nonce in place of timestamp that
eliminates the need for strict time synchronization.
Our scheme has four phases: the registration, login,
authentication and password change. The
communication flow of our authentication protocol
is given in Figure 2.
Table 1: Notations used in the proposed scheme.
⊕
Bit-wise Exclusive-OR operation
||
Bit-wise Concatenation
H(d)
Hash function of d
,
Random nonce
PW
Password chosen by user
t,
Current time recorded by a node
TS
Timestamp for particular user
TID
Temporary User ID
UID
User’s identity
vpw
SHA-4 of user password
x
s
Secret key known to the GW
X
s
Key generated by GW that is also
known to user and sensor nodes.
Figure 1: Typical network setup.
2.1 Registration
There are seven steps of the registration phase R1-
R7 as given below:
R1: A registration interface is launched by a user’s
device (UD) who inputs its ID (UID) and a chosen
password PW. The user computes the Hash of PW
and stores it as vpw.
R2: UD submits its identity UID and vpw to GW
node on a secure channel.
R3: GW stores the secret key
and generates a
random nonce
and computes TID and
as
shown in Figure 2. Then GW stores (TID, vpw,
and TS) in the user database. The flag value for the
user is set to zero.
R4: GW replies the user for a successful
registration with Xs and random number N
0
.
R5: Then UD computes TID as shown in Figure 2
and stores TID and Xs for future use.
R6: GW distributes (TID,
, TS) to those SNs that
are able to provide login interface to the user.
PECCS2013-InternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems
36
Figure 2: Communication flow of the proposed protocol.
R7: A small database is maintained in the sensor
node to store (TID,
, TS) for the user.
2.2 Login Phase
A user needs to login to a sensor node to get sensory
information. The main steps of login phase are:
L1: At time t, a user initiates login and computes A,
which is the hash of vpw in concatenation with t.
L2: The user submits (TID, A, t) to a sensor node.
L3: Upon receiving the login request at time T
0
, the
sensor node checks its database to validate TID. On
validation, the sensor node (SN) retrieves the
corresponding A and computes
as shown in
Figure 2. Otherwise, the login request is rejected.
L4: Sensor node sends (TID, C
k
, T
0
, t) to the GW.
2.3 Authentication Phase
For authentication, GW receives (TID, C
k
, T
0
and t)
from SN at time
. The main steps of this phase are:
A1: GW checks the database whether TID is a valid
user. If TID is not valid then login is declined.
Otherwise, the GW checks whether time t is
PasswordbasedSecureAuthenticationMethodologyforWirelessSensorNetwork
37
recorded against the TID in the database. If t is
already recorded then login is not allowed that will
prevent a replay attack. Otherwise, GW checks the
flag associated with the TID. If the flag is set then
login is rejected due to multiple login attempts. If
not then GW retrieves vpw from the database and
computes
′
and
′
. GW verifies (
=
′
) to
authenticate the SN and user. Otherwise, a reject
message is sent to SN. GW computes V
s
and V
u
, time
t is recorded in the database and the flag is set.
A2: GW sends the accept login message
(Acc_Login,
,
,
) to the sensor node (SN).
A3: At time
, SN receives message from GW,
computes V’
s
and verifies (V
s
=V’
s
) to authenticate
GW node and the user. SN then computes
as
shown in Figure 2.
A4: SN sends (Acc_Login, Y
u
, T
2
) to the UD.
A5: Upon receiving the message at time T
3
, UD
computes V′
u
and Y′
u
. If (Y
u
=Y’
u
) then SN and GW
are authenticated and UD starts retrieving data from
the SN. Otherwise, Acc_Login message is declined.
2.4 Password Change
Main steps of password change phase are as follows:
P1: UD chooses the new password PW
1
and
computes vpw
1
that is the hash of PW
1
.
P2: UD sends the triplet (TID, vpw, vpw
1
) to GW
via a secure channel.
P3: After verification of TID and vpw, GW
generates nonce N
1
and computes TID
1
, X
1s
and
TID′
1
as shown in Figure 2. GW updates TID, vpw,
X
s
and TS.
P4: GW sends success change, Succ_Change (N
1,
X
s
) to the UD.
P5: UD computes TID
1
and updates TID and X
s
.
P6: The GW distributes (TID, TID′
1
, X
1s
, TS
1
) to all
the sensor nodes.
P7: Upon receiving updates, SN checks TID and
computes TID
1
. It also updates TID, X
s
and TS.
3 SECURITY ANALYSIS
We assume that the adversary has the ability to
replay, block or forge any network traffic.
Moreover, it is computationally infeasible to break
the cipher.
3.1 Performance Analysis
We have used computational overhead as a metric to
compare the performance of our proposed scheme
with the Vaidya et al.’s scheme (2009). The
comparison of the computation is presented in Table
2. The number of elements contained in the
messages is not considered for comparison. Our
proposed scheme is slightly computationally
expensive at GW. However, in most of the WSN
applications, the computational capability of GW
and user devices is higher than SN. The one-way
Hash function and the XOR operation are considered
lightweight for these two devices. It means that
without adding any extra computational load for the
SN, our authentication scheme provides higher
security as compared to Vaidya et al.’s method
(2009).
Table 2: Comparison of computational overhead.
Network
Element
Authentication Scheme
Vaidya et al.’s
Scheme
The Proposed Scheme
User (UD)
5
+ 1
5
+ 1
Gateway
Node (GW)
5
+ 3
+
(K+1)
6
+ 3
+
(K+1)
Sensor
Node (SN)
3
+ 2
+
1
3
+ 2
+ 1
Total
13
+ 6
+
(K+2)
14
+ 6
+
(K+2)
The comparison of functional requirements can
also be done easily. One can observe that the Vaidya
et al.’s scheme (2009) does not resist multiple login
and replay attack within a time interval. Their
scheme uses timestamp to avoid replay attack.
However, implementation of strict and safe time
synchronization is difficult. In the case of a shorter
transmission delay interval, a legal user’s login can
also fail. A large transmission delay will lead to
replay attacks. On the other hand, our user
authentication scheme provides better security
features without adding any extra computation at
SN. Our scheme inherently resists replay attack as it
does not rely on network synchronization. The
authentication latency time of our scheme is also
low as it does not need to check the time interval.
3.2 Security Analysis
Our authentication scheme has all the security
features of past schemes such as pseudo-nimity and
resistance to password guessing, impersonation and
replay attacks. It also incorporates mechanism to
remove the flaws in Vaidya et al.’s schemes (2009;
2010).
Replay Attack of Login Message in the Login Step
PECCS2013-InternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems
38
L2: Our scheme can resist replay attack of login
message. When an adversary eavesdrops the login
message (TID, A, t) in the login step L2 and uses it
to impersonate the UD to login the SN in a later
session, our scheme resists this replay attack:
The adversary replays the same previous
message without modifying time, t. GW detects it as
the time, t has already been recorded and the login
will be denied by the GW.
An adversary modifies the message and sends it
to GW. (C
k
= C′
k
) will fail and GW denies the login.
Replay Attack of Accept Login Message in the
Authentication Step A2 and A4: Our scheme can
resist a replay attack on the accept-login message in
the authentication step A2 in the following ways:
While transmitting (Acc_Login,
,
,
) from
GW to SN, the malicious party can intercept the
message before forwarding it.
In the next session when a legitimate SN sends
(TID,
,
, t) to GW, the malicious party
intercepts and drops that message to replay the
captured Acc_Login message to SN pretending itself
a legal GW.
Since A′ in V
s
i.e. H(
|| A′||
||
) is different
from A in V′
s
, the verification of (V
s
=V′
s
) will fail
and the login will be rejected by SN.
Alternatively, the malicious party can modify the
Acc_Login message by replacing T
1
with T
1e
and
send the message to SN. Since
in
is different
from T
1e
in V′
s
, the verification of (V
s
=V′
s
) will fail
and the login will be rejected by SN.
Similarly, the replay attack of accept login message
in authentication step A4 is resisted by our method.
Replay Attack of Messages in Login Step L2 and
Authentication Step A2: The adversary eavesdrops
the login message in login step L2 and (Acc_Login,
,
,
) in the authentication phase step A2.
The adversary replays the message (TID, A, t) to
the sensor node. SN checks that the TID is a valid
user and computes C
ke
at time T
0e
.
The adversary blocks the message sent from the
SN to GW in the login step L4 preventing the GW
from receiving this message.
The adversary replays (Acc_Login,
,
,
)
message in the authentication step A2 to SN.
SN receives message at T
2e
and computes
V
s
=H(
||A||
||
) that is different from
V′
s
=H(
||A′||
||
) in the replayed message. In
this way, login will be denied in the sensor node.
Forgery Attack with Node Capture Attack: Our user
authentication scheme resists a forgery attack in two
ways as given here.
After capturing SN, the adversary cannot
compute C
k
since A is not known or the verification
of (C
k
= C′
k
) will fail and the login will be denied.
Gateway Bypass Attack Due to Node-capture
Attack: If a user is allowed to access data from
sensor node directly without GW node’s notice then
the impact of “node compromise” attack is very
high. Our scheme resists gateway bypass attack.
Adversary can compute
=H(
||A||
) in
authentication step A3 but cannot compute V
u
since
is not known. In this way, the gateway bypass
attack is prevented in our scheme.
Many logged in user with same login ID threat: If a
valid user shares its TID and password with another
user, the other user can generate the login message
and gain access to the login node. Our scheme can
resist it as given below.
GW records the time t against the TID and the
flag of TID is changed from zero to one indicating
that a user with this TID has logged in to the
network. GW will decline such login requests.
Mutual Authentication: The proposed scheme can
provide mutual authentication in the following way:
UD
i
is authenticated to GW by A = H(vpw||t)
GW is authenticated to UD
i
by V
u
UD
i
is authenticated to SN by TID
SN is authenticated to UD
i
by Y
u
SN is authenticated to GW by C
k
GW is authenticated to SN by V
s
4 PROTOCOL MODELING
We are evaluating the performance of our scheme
under different attack scenarios. We have modelled
a WSN using SystemC (2011). The SystemC based
module architecture for our user authentication
methodology is shown in Figure 3.
The WSN model consists of SystemC modules
communicating through channels. Each module
contains at least one process. Processes are activated
according to a sensitivity list defined statically or
dynamically. A module is made of communication
and software models. The communication model
emphasizes the communication between sensor,
gateway, intruder and user modules. Software model
consists of C++ processes representing sensor,
gateway, user and other WSN components.
The user module generates the registration
packet from the data provided by the user and sends
the packet to gateway modules. The packet size used
in our scheme is of 382 bytes length. The hash
PasswordbasedSecureAuthenticationMethodologyforWirelessSensorNetwork
39
function SHA-512 is used to generate the message
digest in our scheme. However, 160 bit SHA-1 hash
function can also be used to reduce the memory
requirement for gateway (GW) and sensor modules.
Figure 3: SystemC architecture of our protocol.
The authentication latency of our scheme is a
function of the number of users, encryption cipher
and channel transmission rate among the modules.
We executed the simulation 20 times and observed
that the average authentication latency of our
scheme is 0.29 seconds (for one user). The effect of
simultaneous multiple users’ login requests on
authentication latency is also being investigated.
Replay, node capture, gateway bypass and multiple
login attacks are further investigated by simulating
our authentication protocol. It is verified that the
protocol is resistant against all the above attacks that
re-affirms our claims of Section 3. The main focus
of our scheme is to provide application layer
security. However, the security can further be
enhanced by incorporating IEEE 802.15.4 (2003)
specification at MAC sub-layer for all the phases of
the proposed authentication protocol.
Finally, a memory requirement comparison is
made between the Vaidya et al.’s scheme (2009) and
our protocol. The user data storage requirements for
both schemes are presented in Table 3. The storage
requirement for sensor node in our scheme is
slightly higher than Vaidya et al.’s scheme (2009).
However, our protocol is resistant to replay as well
as gateway bypass attacks.
Table 3: Comparison of storage overhead.
Storage Overhead per User (bits)
UD GW SN
Vaidya et al.’s scheme (2009) 2304 1600 1056
The Proposed scheme 2304 1601 1088
5 CONCLUSIONS
In this paper, we have proposed a robust user
authentication scheme, which is an improved
password-based authentication method. We have
identified that Vaidya et al.’s scheme is subject to
several security flaws (2009). To overcome these
flaws, we have proposed an improved scheme that
retains all the advantages of past user authentication
schemes. Our proposed scheme resists replay, GW
node bypass and many logged in user attacks. The
scheme provides mutual authentication and resists
replay attack inherently and does not require any
network synchronization. We have modeled our
protocol using SystemC and verified different attack
scenarios. In our scheme, we have assumed that the
database is securely stored in GW node and failing
to meet this condition may make our scheme
vulnerable to stolen verifier attack. Again none of
the past schemes provides an inherent method to
detect a compromised node.
REFERENCES
Benenson, Z., Gartner, F., & Kesdogan, D., (2004). User
authentication in sensor networks, In Proc. Workshop
on Sensor Networks. Informatik.
Benenson, Z., Gedicke N., & Raivio, O., (2005). Realizing
robust user authentication in sensor networks, In Proc.
Workshop Real-World Wireless Sensor Networks,
Stockholm.
Das, M. L., (2009). Two-factor user authentication in
wireless sensor networks, IEEE Trans. Wireless
Communication, 8, 1086-1090.
IEEE Standard, 802.15.4-2003, (2003). Wireless medium
access control and physical layer specifications for
low-rate wireless personal area networks.
Khan, M. K., & Alghathbar, K., (2010). Cryptanalysis and
security improvements of two-factor user
authentication in wireless sensor networks, Sensors,
10(3) 2450-2459.
Ko, L. C., (2008). A novel dynamic user authentication
scheme for wireless sensor networks," In Proc. IEEE
ISWCS, 608-612, Reykjavik Iceland.
Lee, T. H., (2008). Simple dynamic user authentication
protocols for wireless sensor networks, IEEE Sensor
Communication, 43, 657-660.
Sastry, N., & Wagner, D., (2004). Security considerations
for IEEE 802.15.4 networks, In Proc. ACM Workshop
on Wireless Security, 32-42.
SystemC. The language for system-level modeling, design
& verification, IEEE Std. 1666-2011. Retrieved from
http://www.accellera.org/downloads/standards/System
C.
Tseng, H. R., Jan R. H., & Yang, W., (2007). An
improved dynamic user authentication scheme for
Gateway
Module
Sensor Module
Sensor Module
Sensor Module
User Module
Authentication Protocol
Specification
PECCS2013-InternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems
40
wireless sensor networks, In Proc. IEEE GlobeCom.
986–990.
Vaidya, B., Makrakis, D., & Mouftah, H., (2010).
Improved two-factor user authentication in wireless
sensor networks, In Proc. Int. Wks. Network
Assurance & Security Services in Ubiq. Environments,
600-605.
Vaidya, B., Rodrigues, J. J., & Park, J. H., (2009). User
authentication schemes with pseudonymity for
ubiquitous sensor network in NGN, Int. Journal of
Communication Systems, 23, 1201-1222.
Wong, K. H., Zheng, Y, Cao, J., & Wang, S., (2006). A
dynamic user authentication scheme for wireless
sensor networks, In Proc. IEEE SUTC, 1, 318–327.
Zhou, X., Xiong, Y., Miao F., & Li, M., (2011). A new
dynamic user authentication scheme using smart cards
for wireless sensor network, In Proc. IEEE Int. Conf.
Computing, Control and Industrial Engineering, 1-4.
PasswordbasedSecureAuthenticationMethodologyforWirelessSensorNetwork
41
