Secret Sharing Scheme and Key Sharing Scheme Suitable for
Clustered Sensor Networks
Shinichi Goto
1
, Keiichi Iwamura
1
, Yuji Suga
2
and Kitahiro Kaneda
3
1
Department of Electrical Engineering, Tokyo University of Science, 6-3-1 Niijuku, Katsushika-ku, Tokyo, Japan
2
Internet Initiative Japan, 2-10-2 Fujimi, Chiyoda-ku, Tokyo, Japan
3
Institute of Document Analysis and Knowledge Science, Osaka Prefecture University, 1-1 Naka-ku, Sakai, Osaka, Japan
Keywords: Sensor Network, Key Sharing, Secret Sharing Scheme, Node Analysis, Storage Capacity Reduction.
Abstract: A clustered sensor network collects two or more nodes and makes a cluster from them. Generally a cluster
head (CH) has all the encryption keys of the nodes in a cluster and performs encrypted communication.
However, this type of network has a problem in that the keys are revealed if the CH is analyzed every time
after key sharing. Moreover, when all the nodes are set to CH, another problem arises, which is the need for
large storage for holding the keys of all the nodes in a cluster. In this paper, we propose the first key sharing
scheme that carries out key sharing with the CH and each node in a cluster and that realizes information
theoretical security using a secret sharing scheme, even if the CH is analyzed except for the time of
encryption communication. Next, we propose the second key sharing scheme in which additional storage for
the CH for saving the keys of all the nodes in a cluster is not needed. In order to realize it, the secret sharing
scheme is improved and the security is evaluated. In addition, we present the third key sharing scheme in
which none of the keys are revealed at all even if CH or all the child nodes are analyzed.
1 INTRODUCTION
Wireless sensor networks are known as ad hoc
networks that consist only of sensor nodes
connectable by radio. Generally, the sensor nodes
send various data such as the temperature, humidity,
etc., to a base station (BS) directly or indirectly
through multi-hop routing. Sensor networks can be
used for various purposes, from military to
noncommercial uses, and are expected to become
the next-generation communication infrastructure.
However, sensor nodes are not tamper resistant
and do not have high computational resources owing
to cost factors. Therefore, it is difficult to perform
complicated calculations like public key encryption.
Additionally, since the memory capacity of a sensor
node is small, the amount of data it can hold is
limited. Moreover, sensor nodes have a low-capacity
battery. Therefore, efficient energy consumption is
an important consideration point. In addition, since
sensor nodes are often placed outdoors to monitor
the information in the field, an attacker can easily
steal and analyze the nodes to obtain sensitive
information. In contrast, a BS is generally managed
securely and has enough electrical power and
computational resources.
The main research on sensor networks is how to
decrease the energy consumption of sensor nodes,
and there are many results on efficient energy
consumption. One of these is LEACH (Low Energy
Adaptive Clustering Hierarchy) (Heinzelman et al.,
2000), which has been proposed as a clustered
sensor network. A clustered sensor network makes a
group of nodes called a cluster and sets one of these
nodes as the cluster head (CH). The nodes in each
cluster send the sensing data to the CH. The CH then
forwards the data to the BS. What LEACH does is to
change the CHs in turn so that the clusters are
independently composed without the need for a BS.
In LEACH, when all the nodes are periodically set
to CH, the deviation of the energy consumption of
all the nodes is equalized, allowing the life of the
network to be extended.
However, the original LEACH protocol did not
consider the issue of security. Thus, some new
LEACH protocols that realize security using
common key encryption, such as SecLEACH
(Oliveira et al., 2007), MS-LEACH (Qiang et al.,
2009), etc., were proposed.
172
Goto, S., Iwamura, K., Suga, Y. and Kaneda, K.
Secret Sharing Scheme and Key Sharing Scheme Suitable for Clustered Sensor Networks.
DOI: 10.5220/0005999601720180
In Proceedings of the 13th International Joint Conference on e-Business and Telecommunications (ICETE 2016) - Volume 6: WINSYS, pages 172-180
ISBN: 978-989-758-196-0
Copyright
c
2016 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
However, almost all of the key sharing schemes
in a clustered sensor network have the same
problem: since the CH has the encryption keys of all
the nodes in the cluster to perform encrypted
communication, if the CH is analyzed, these
encryption keys are revealed. Moreover, when all
the nodes are set to the CH like in LEACH, another
problem arises, which is the need for big storage to
hold the keys of all the nodes in the cluster.
On the other hand, some key sharing schemes
use (k,n) secret sharing scheme. The secret sharing
scheme makes n shares from a secret, and the secret
can be restored from k (k
n) shares. In the key
sharing scheme, when CH does not have any keys, it
restores the encryption key by receiving shares from
k neighboring nodes (Bertier et al., 2010). However
in these schemes, the encryption key is leaked either
by the analysis of k neighboring nodes or k
communication paths, both of which are smaller
than the total number of child nodes. Therefore,
some schemes, e.g., that proposed in (Yiying et al.,
2013) use a public key cryptosystem to hide the
shares. However, because of the computational
complexity of a public key cryptosystem, the energy
consumption of a node is very large.
In this paper, we propose the following three
kind of key sharing schemes using secret sharing
schemes.
(1) The first scheme realizes that, even if the nodes
including the CH are analyzed, the CH does not
at all reveal the key between the nodes that are
not analyzed. This means that this scheme
realizes information theoretical security on key
analysis.
(2) The second scheme is a key sharing scheme in
which the CH does not need to save the keys or
the shares on all the nodes in a cluster, but
manages only its own key. Realization of this
scheme requires the secret sharing scheme to be
improved. This scheme achieves computational
security.
(3) The third scheme is a key sharing scheme that
does not leak the key at all even if CH or all of
the child nodes are analyzed. This scheme can
either select information theoretical security or
computational security.
The first scheme is recommended if the user
wants to focus on information theoretical security
against CH analysis and if the storage capacity of the
nodes is sufficient to hold the keys. The second
scheme is recommended for Internet of Things (IoT)
device that requires fewer calculation and memory
resources. The third scheme is suitable for group key
sharing which is used as a common key within a
cluster.
The remainder of this paper is organized as
follows: In Section II, we explain LEACH and
discuss existing research studies on LEACH with
security. Section III describes the first scheme using
an existing secret sharing scheme. Section IV
describes the improved secret sharing scheme and
the second scheme. Section V presents the third
scheme and its variations. Finally, in Section VI, we
describe the performance evaluation.
2 EXISTING RESEARCH
STUDIES
2.1 Leach
LEACH is a protocol that selects a node that, in turn,
becomes the CH and averages the energy
consumption of all nodes to extend the life of
clustered sensor networks.
LEACH has two communication phases: a setup
phase and a steady-state phase. In the setup phase,
LEACH uses a random number to choose a CH in a
cluster. The chosen node then broadcasts a message
that it has become the CH. The nodes choose the
nearest CH and send a message that they have
become child nodes. The CH then sends a time
division multiple access (TDMA) schedule for the
steady-state phase to the child nodes. In the steady-
state phase, the nodes send the sensing data to the
CH according to the TDMA schedule. The CH
compresses the data received from multiple child
nodes and transmits the combined data to the BS.
2.2 SecLEACH
In SecLEACH, an administrator sets some element
keys to each node before use at random from a key
pool, which is a set of element keys. Each element
key has a key ID. SecLEACH performs key sharing
in the setup phase as follows:
1. The CH announces the key IDs to the child
nodes.
2. Each child node selects a key ID(s) that is (are)
common to the CH.
3. Each child node transmits the common key ID(s)
to the CH.
4. Each child node and the CH generate and save
the common key, which was generated by the
common element key(s).
Secret Sharing Scheme and Key Sharing Scheme Suitable for Clustered Sensor Networks
173
SecLEACH can determine a common key by
knowing only the ID of the element keys. However,
SecLEACH has some problems. First, the key
sharing between the CH and a node is probabilistic
because there is not necessarily a common element
key. If a child node does not have a common key,
the node cannot communicate with the CH. Second,
if an attacker steals and analyzes nodes, the keys of
the other nodes may be analyzed. That is,
SecLEACH is weak to node analysis. If the number
of element keys that are saved is increased, analysis
will become difficult and the problem of requiring
more memory for the element keys arises.
In addition, all the keys of the nodes in a cluster
are revealed if the CH is analyzed every time after
key sharing.
2.3 MS-LEACH
MS-LEACH uses the Localized Encryption and
Authentication Protocol (LEAP) to obtain a common
key.
All the nodes have a function to generate
pseudo-random number. The procedure for
obtaining a common key is as follows:
1. An administrator makes an initial key K and sets
it to all nodes. A child node u makes a master
key Ku = E
K
(u). E
K
(u) is a pseudo-random
function that uses as input K and node ID u.
2. The CH and a child node transmit their own ID
to each other.
3. A child node u makes a common key K
UV
=
E
Ku
(ch) using the pseudo-random function, with
Ku and the CH ID as input.
4. The CH makes the master key Ku using the
pseudo-random function, with K and a child
node’s ID as input. After that, the CH makes a
common key K
UV
= E
Ku
(ch) using the master key
Ku and its own ID.
5. All the nodes delete the initial key K after
making a common key.
When MS-LEACH is used, the CH can share the
keys with all nodes certainly. In MS-LEACH, even
if an attacker steals and analyzes a node after the
deletion of the initial key, the other links remain
secure. However, all common keys are leaked if an
attacker obtains an initial key K, since MS-LEACH
generates the entire key from the initial key. In
addition, MS-LEACH does not realize information
theoretical security since it makes the common key
using a pseudo-random function.
All the keys of the nodes in a cluster are revealed
if the CH is analyzed every time after key sharing.
2.4 SSKM: Secret Sharing-based Key
Management
To keep secure channel for delivering shares, SSKM
(Yiying et al., 2013) adopt the discrete logarithm in
the finite field and DDH difficulty assumption.
Therefore, this scheme requires a large amount of
communication and computational complexity,
because the discrete logarithm is calculated on a
large finite field.
This scheme shares one cluster key (group key)
between all the nodes.
[Initial phase]
Assume that there are 1 clusters, and each
cluster has a cluster head and n (n k) member
nodes. In this phase, BS sets the parameters for key
sharing.
1. BS chooses two big primes 1 and 1; let =
21 + 1 and = 21 + 1, = ; it is
computationally intractable to solve the factor
without , . Meanwhile, BS selects a generator
( [
1/2
,]) and another prime ( > ).
And then, BS broadcasts the three triple (, ,
) to sensors in the network.
2. BS randomly and uniformly chooses a
polynomial () of (k 1)-degree for each
cluster as follows:
() =  +
+
+⋯+


3. BS independently selects a session key
CH
from
GF() in the finite field and hides the session
key with secret
CH
, namely Z
CH
=K
CH
+S
CH
.
[Cluster key management]
1. CH chooses xch randomly which relatively
primes with p-1 and q-1, and CH sends it to BS.
Then, BS counts out y
CH
=
xch
and sends (ID
CH
,
y
CH
) to sensor node in cluster; meanwhile,
sensor node picks xi randomly which relatively
primes with p-1 and q-1, computes yi=
xi
mod
N, and then sends (IDi, yi) to the BS. The BS
ensures that if IDi
IDj, there should be no
yi=yj; otherwise it reselects until success.
Furthermore, the BS utilizes CH's ID
CH
and
members' IDi (i=1,...,k) to count out the share
f
CH
(ID
CH
) and f
CH
(IDi), respectively.
2. The CH selects a group of users  = {ID1,…,
IDk}, while BS unicasts (ID
CH
, f
CH
(ID
CH
)(yi)
xch
mod N) to sensor node in the cluster and
sends (IDi, f
CH
(IDi)(y
CH
)
xi
mod ) to CH.
3. BS sends Z
CH
not to leak the K
CH
.
[Secret recovery]
WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems
174
Depending on the received information from BS,
public generator, node’s private key , and CH’s
own key
CH
, cluster head and members can obtain
their share through the following formulas:
)(
)(
)()(
)(
)(
)()(
iCH
x
i
x
chiCH
CHCH
x
ch
x
iCHCH
IDf
y
yIDf
IDf
y
yIDf
ch
i
i
ch
=
=
3 FIRST SCHEME
3.1 A Fast Secret Sharing Scheme
using XOR
Shamir’s secret sharing scheme is a typical scheme.
However, since this scheme needs a polynomial
operation, it is complicated to process in a sensor
node. Therefore, we selected a fast secret sharing
scheme (Kurihara et al., 2008) using just an
exclusive OR (XOR) operation. This scheme has
minimal processing requirements and information
theoretical security. The details of this scheme are
omitted owing to page restrictions.
3.2 Application to LEACH
Existing key sharing schemes (Bertier et al., 2010)
distributes n shares to n nodes and collects any k (k
n) out of n to recover the encryption key.
Therefore, the key is analyzed if k communication
paths are eavesdropped. In contrast, since the
proposed scheme sets threshold k to be larger than
the number of target child nodes, the key does not
leak even if all communication paths from the child
nodes are eavesdropped. Only the targeting node
that performs encryption communication holds the
multiple shares or the key itself to realize the key
recovery.
First, we put the following assumptions:
A cluster contains m + 1 nodes whose IDs are
named ID
1
, ID
2
, ID
3
, …, ID
m
, and ID
m+1
. The
number of child nodes is m and CH is 1.
Each node stores its own ID, the unique key K
i
,
and the link key L
i
in advance. K
i
is different for
each node. L
i
is used as a common key for the
encrypted communication between the CH and a
node. L
i
is independently selected from uniform
random number.
The BS knows the aforementioned information
about all nodes, and communication using the
unique key K
i
between the BS and each node is
secure.
The proposed scheme can be applied independently
to every cluster or CH even if there are two or more
clusters or CHs. Our schemes are performed in the
setup phase after selecting the CH. In the first
scheme, we set k=2. Since each child node performs
a one-to-one key sharing with the CH, the number of
the targeting child node is 1 for one key. We can
select (n,k)=(3,2) and a child node that has W
i1
and
W
i2
without storing the link key, but the following is
easier.
In our schemes, although BS sends encrypted
shares to each child node directly to keep the
explanation simple, the shares can be sent via CH.
Since CH does not know each unique key, CH
cannot decrypt it. The first scheme is performed as
follows:
[Distribution]
1. The CH transmits the IDs of the child nodes in
the cluster to the BS.
2. The BS sets n=k=2 and performs the secret
sharing scheme in 3.1 for each link key L
i
as a
secret information independently, and it
generates two shares, W
i0
and W
i1
.
3. The BS encrypts W
i0
(i=1,…,m) with the unique
key of the CH and sends them to the CH. The
CH decrypts W
i0
(i=1,…,m) and saves them in
association with ID
i
.
4. The BS encrypts W
i1
with each unique key of ID
i
and sends it to ID
i
. Each ID
i
decrypts W
i1
and
saves it.
[Key sharing and encrypted communication]
1. ID
i
sends the following information to CH. E
Li
(f
i
)
is encrypted data of the sensing data f
i
using the
link key L
i
.
(ID
i
, W
i1
, E
Li
(f
i
))
2. The CH restores the link key L
i
from W
i0
and W
i1
.
3. The CH obtains the sensing data f
i
by decrypting
E
Li
(f
i
) using L
i
.
4. The CH deletes the link key L
i
.
3.3 Security Assessment
(1) Security Against CH Analysis
The CH has only one share on one key. Therefore,
since k=2, no link key is revealed even if the CH is
analyzed except for the time of encryption
communication. This scheme has information
theoretical security based on the secret sharing
scheme in 3.1. Since key sharing between the CH
Secret Sharing Scheme and Key Sharing Scheme Suitable for Clustered Sensor Networks
175
and each node is independently set, even if two or
more CHs are analyzed, the security does not change.
However, at the time of encryption communication,
since CH restores the link key to communicate with
a child node, the link key is revealed. However, a
key sharing scheme equal to an attack that combines
CH analysis and information to CH to generate a key,
does not exist as far as the authors know. In addition,
since the CH in other schemes contains all the keys
of the child nodes, all these keys are leaked every
time the CH is analyzed.
(2) Security Against Child Node Analysis
Each child node has its own link key L
i
. Therefore,
the link key is leaked if a child node is analyzed.
However, L
i
is set independently, which means that
the link keys of the other nodes are not revealed at
all; thus, this scheme realizes information theoretical
security also in regard to child node analysis. In
contrast, as described previously, when many child
nodes are analyzed in SecLEACH, or when a child
node is analyzed before eliminating the initial key in
MS-LEACH, the other link keys are revealed.
(3) Security Against CH and Child Node Analysis
The first scheme has the same security as the child
node analysis, even if both the CH and child nodes
are analyzed. This is because each link key L
i
is set
up independently, and the number of shares that the
CH has is one for one key.
(4) Security Against Eavesdropping on
Communication Paths
Although W
i1
is sent as it is in the key sharing and
encrypted communication phase, since W
i0
does not
appear in the communication path, information
theoretical security is realized. SecLEACH also
realizes information theoretical security since it
sends only the key ID, whereas MS-LEACH realizes
computational security.
4 SECOND SCHEME
In the first scheme, the CH needs to save all the
shares of the child nodes. Therefore, we propose the
second scheme in which the CH does not need to
save these shares, but only manages its own key.
However, the second scheme cannot be realized by
applying the secret sharing scheme in 3.1 as it is.
Therefore, we apply asymmetric reduction
(Takahashi et al., 2014) to the scheme in 3.1.
Asymmetric reduction reduces the shares of each
server non-uniformly, in contrast to the ramp
scheme, which reduces the shares in each server
uniformly. In other words, the asymmetric secret
sharing scheme can set the shares on the server up to
k-1 to 0 (it manages only a key). The asymmetric
reduction in (Takahashi et al., 2014) was applied to
Shamir’s secret sharing scheme and has been proven
to have computational security, which depends on
the security of the pseudo-random number used.
Therefore, we show the asymmetric secret sharing
scheme of 3.1 as follows.
4.1 Asymmetric Secret Sharing Scheme
using XOR
In the following scheme, secret information S has an
ID named IS, and the size of the secret key is (n-1)d
bits. ‘’ is XOR, and ‘||’ is the connection of the bit
sequence.
[Distribution]
1. The dealer arbitrarily selects t (1
t
k-1)
servers in n servers and calls them key servers.
The ID of the key servers is i (1
i
t). The
dealer and the key server have a pseudo-
random number generator.
2. The dealer sets key
i
to each key server i.
3. The dealer generates a pseudo-random number
qi=E
keyi
(IS) of (n-1)d bits from the ID of S using
the key
i
of each server, and it is considered as the
share of the key server.
4. The dealer divides
for every d bits and makes
(
,
)
(0 − 2). He divides S for every d
bits and makes partially secret information items
(
,
,…,

). We assume
=0.
=
(,)
||
(,)
|| ||
(,)
=
||
|| ||

, 
0
5. The dealer assumes the following n(k-1)-1
random numbers.
,
,⋯,

,
,⋯,

,

,⋯,

,⋯,


6. The dealer generates (k-1)n-(n-1)t-1 random
numbers and arbitrarily assigns them to the
aforementioned
∙
.
7. The dealer calculates the remaining (n-1)t
random numbers of
∙
from
(
,
)
,⋯,
(
,
)
,⋯,
(
,
)
such that the
following equation is realized:
(
,
)
=

⨁⊕


∙
(1≤≤,0≤≤−2)
8. The dealer generates the partial shares as follows
using the assigned and calculated
∙
:
(,)
=

⨁⊕


∙
WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems
176
(t+1≤≤1,0≤≤2)
9. The dealer generates the remaining shares
by
connecting each partial share and distributes
to the server except (1 ) . The server
stores the shares. We call the server data server.
=
(,)
||
(,)
|| ||
(,)
(
0,1
(

)
)
[Restoration]
When we restore S, we select k servers from n servers
and send the ID of S to the servers. When we select a
key server i, the server generates and sends the share
=

(IS). When we select a data server, the
server sends the stored share. S is restored by the
same restoration process as in 3.1 using these k shares.
4.2 Application to LEACH
The second scheme has the same premise as the first
scheme except that CH has a pseudo-random
number generator.
[Distribution]
1. The CH transmits the IDs of the child nodes in
the cluster to the BS.
2. The BS sets n=k=2, t=1, and performs the secret
sharing scheme shown in 4.1 for each link key L
i
as the secret information independently, and it
then generates two shares, W
i0
and W
i1
. Here, the
CH is selected as the key server. Thus, W
i0
=
and W
i1
=Wi in 4.1. IS is the ID of each node.
3. The BS encrypts W
i1
with each unique key of ID
i
and sends it to ID
i
(i = 1, ···, m). Each ID
i
decrypts W
i1
and saves it.
[Key sharing and encrypted communication]
1. ID
i
sends the following information to CH. E
Li
(f
i
)
is the sensing data f
i
that was encrypted using the
link key L
i
.
(ID
i
, W
i1
, E
Li
(f
i
))
2. The CH generates W
i0
using the pseudo-random
number generator from ID
i
and key
CH
, and
restores L
i
from W
i0
and W
i1
.
3. The CH obtains the sensing data f
i
by decrypting
E
Li
(f
i
) using L
i
.
4. The CH deletes the link key L
i
.
4.3 Security Assessment
The scheme in 4.1 has computational security,
depending on the used pseudo-random number, but
the proof is omitted because of page restrictions. The
security of the second scheme is shown as follows:
4.3.1 Security against CH Analysis
When the CH is analyzed, a key for generating the
shares
will be revealed. This is equivalent to the
case in which all the shares in CH are revealed in the
first scheme. In the second scheme, since each link
key is independently set up as in the first scheme, no
link key is revealed from just one share. This means
that the second scheme has information theoretical
security similar to the first scheme on CH analysis.
4.3.2 Security against Child Node Analysis
The link key of a child node is also revealed in the
second scheme. In this scheme, W
i0
=
and
W
i1
=L
i
⨁
. When the node ID
i
is analyzed,
is
known from W
i1
and L
i
. If the pseudo-random
numbers that are used are vulnerable,
may be
obtained from some
. However, even in such a case,
if W
j1
in node ID
j
is not obtained, L
j
is not revealed,
since the analysis of
= W
j0
does not contain L
j
. That
is, the key of a node that is not analyzed is not
revealed at all, even if many child nodes may be
analyzed, and the second scheme has information
theoretical security also on child node analysis.
4.3.3 Security against CH and Child Node
Analysis
The second scheme as well as the first scheme has the
same security as far as child node analysis is concerned.
4.3.4 Security against Eavesdropping on
Communication Paths
In the key sharing and encrypted communication
phase, W
i1
is sent as it is from a child node. If
is
known because of the vulnerability of the pseudo-
random numbers, L
i
may be revealed, since W
i1
contains L
i
. Therefore, in this case, it is accepted that
this scheme depends on the computational security
of the pseudo-random numbers that were used.
5 THIRD SCHEME AND ITS
VARIATIONS
5.1 Application to LEACH
The third scheme has the same premise as the first
scheme except that each node does not have a link
key. In this scheme, it is assumed that threshold
k=m+1 and m>1. This means even if all the child
Secret Sharing Scheme and Key Sharing Scheme Suitable for Clustered Sensor Networks
177
nodes are analyzed, none of the keys are revealed at
all. In addition, we assume that nodes up to u (0≤
≤2) may be unable to communicate.
[Distribution]
1. CH informs the IDs of the child nodes in the
cluster to BS.
2. BS decides each link key L
i
for node ID
i
(i=1, ・・・,m), sets n=m+u+2 and k=m+1, and
performs the secret sharing scheme in 3.1 for
each Li as the secret information independently,
and BS generates n shares W
ij
(j=1,・・・,n) for
each ID
i
.
3. BS encrypts the W
ij
(i=1, ・・・,m) with each
unique key of ID
i
and send them to ID
i
,
whereupon ID
i
decrypts and saves them.
4. BS encrypts the W
i,m+1
W
i,m+u+2
(i=1, ・・・,m)
with the unique key of CH and sends it to CH.
[Key sharing and encrypted communication]
1. When ID
i
sends the sensing data f
i
to CH, ID
i
broadcasts its own ID to all nodes.
2. CH and nodes ID
j
except ID
i
broadcasts a share
W
ij
. CH sends shares until m when the distributed
shares are less than m form W
i,m+3
W
i,m+u+2.
3. ID
i
restores the link key L
i
from the distributed m
shares and the share which only ID
i
has. ID
i
sends E
Li
(f
i
), which is encrypted sensing data f
i
with the link key L
i
.
4. CH restores the link key L
i
from the distributed
m shares and the remaining share W
i,m+2
which is
only contained by CH.
5. CH gets the sensing data f
i
by decrypting E
Li
(f
i
)
using L
i
.
6. After communication, CH and ID
i
delete the link
key.
If this scheme selects the secret sharing scheme
in 4.1 instead of that of 3.1, and all child node are set
as key server (t=m), none of the child nodes would
have any share; instead, they would simply be
managing a key.
We show a variation suitable for group key
sharing. It changes the following portions of the
third scheme.
[Distribution]
2. BS decides a group key L for a cluster, sets
n=m+u+2 and k=m+2, and performs the secret
sharing scheme in 3.1 for L as the secret
information, and BS generates n shares W
j
(j=1,・・・,n).
3. BS encrypts each W
i
(i=1,・・・,m) and W
m+2
with
each unique key of ID
i
and send them to ID
i
.
ID
i
decrypts and saves them.
4. BS encrypts the W
m+1
W
m+u+2
with unique key
of CH and send them to CH.
[Key sharing]
2. All nodes including CH broadcasts a share of
those other than W
m+2
. CH sends shares until
m+1 when the distributed shares are fewer than
m+1 from W
m+3
W
m+u+2
.
3. All nodes including CH restores the link key L
from the distributed m+1 shares and W
m+2
.
This variation is named the third dash scheme.
This scheme can also select the secret sharing
scheme in 4.1 instead of that of 3.1
5.2 Security Assessment
5.2.1 Security against CH Analysis
In the third and dash schemes, CH has u+2 (0≤
−2) shares on one key. Therefore, none of the
link keys are revealed even if CH is analyzed. These
schemes have information theoretical security based
on the secret sharing scheme in 3.1. Since key
sharing between CH and each node occurs
independently, even if two or more CH(s) are
analyzed, the security does not change.
5.2.2 Security against Child Node Analysis
In the third scheme, each child node has one share
on one key, the number of child nodes is , and
k=m+1. Thus, none of the link keys are leaked if all
the child nodes are analyzed.
In the dash scheme, each child node has one
share W
j
for every node and a common share W
m+2
,
and k=m+2. Thus, any link key is not leaked if all
the child nodes are analysed since the known share
is m+1.
5.2.3 Security against CH and Child Node
Analysis
The third scheme has robustness against analysis of
CH and −u−2 nodes, since CH has u+2 shares.
The dash scheme has robustness against analysis
of CH and −u−1 nodes.
WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems
178
5.2.4 Security against Eavesdropping on
Communication Path
Since the number of shares on communication path
is less than threshold k, information theoretical
security is realized in the third and dash schemes.
6 PERFORMANCE
EVALUATION
We consider the case in which the number of child
nodes is m and CH is 1 within a cluster. For simplicity,
the length of each ID and each key is denoted by L,
and the length of the sensing information is expressed
by H. It is presupposed that one node communicates c
times in one round. We evaluate the case in which all
the child nodes share a key with the CH and one child
node carries out encryption communication with the
CH. In this case, the CH has already known the ID of
all the child nodes, and the child node has already
known the ID of the CH. In this evaluation, the BS is
not a candidate for evaluation since it has enough
electric power and computational resources; we thus
evaluate only the CH and a child node. We compared
SecLEACH and MS-LEACH with our schemes in the
key sharing phase (“Distribution” in our schemes) and
communication phase (“Key sharing and encrypted
communication” in our schemes). We do not include
SSKM in 2.4 in the comparison, since SSKM is
clearly inefficient compared with other schemes.
6.1 Evaluation of the Communication
Traffic
In SecLEACH, the number of keys that a node has is
set to a, and the average number of keys that is in
agreement with CH is set to b. From the viewpoint
of communication traffic, MS-LEACH is the best. If
a>m and b>c, the first and second schemes are
better than SecLEACH. Although the third scheme
is worst, if H is larger than
mL enough, H becomes
dominant, and the difference becomes small.
Table 1: Comparison of data traffic.
Key sharing Communication
CH
Child
node
Child node
SecLEACH aL bL c(L+H)
MS-LEACH 0 0 c(L+H)
1-st scheme mL 0 c(2L+H)
2-nd scheme mL 0 c(2L+H)
3-rd scheme mL 0 c{(m+1)L+H}
3-rd’ scheme mL 0 (m+1)L+cH
6.2 Evaluation of the Memory Capacity
In LEACH, since all the nodes turn into CH, the
memory size of the largest one serves as the amount
of memory of all the nodes. The CH, except in the
second and third dash scheme, stores the pair of IDs
of child nodes and the keys, in addition to the
memory capacity. Therefore, the second scheme is
the best, and the second and third dash schemes do
not need additional memory depending on the
number of nodes.
Table 2: Comparison of the amount of memory.
Child node CH
SecLEACH 2aL 2aL+2mL
MS-LEACH 2L 2L+2mL
1-st scheme 4L 4L+2mL
2-nd scheme 4L 4L
3-rd scheme 2(1+m)L (2+3m)L
3-rd’ scheme 4L (4+u)L
6.3 Evaluation of Computational
Complexity
We counted the number of times of encryption and
decryption since their computational complexity is
larger than that of the other operations. Therefore,
the computational complexity of the comparison of
the key ID in SecLEACH and of the XOR operation
in our schemes was set to 0.
From the viewpoint of the computational
complexity, SecLEACH is the best. The amount of
calculation of MS-LEACH is larger than that of the
first scheme and the third dash scheme, and if c<2m,
it is larger than that of the second scheme, although
the second scheme adds the c encryption of the ID in
order to make a share.
Table 3: Comparison of calculation.
Key sharing Communication
CH
Child
node
CH
Sum
of
child
node
SecLEACH 0 0 cH cH
MS-LEACH 2mL 2L cH cH
1-st scheme mL L cH cH
2-nd scheme 0 L c(L+H) cH
3-rd scheme 2mL ml cH cH
3-rd’ scheme (2+u)L 2L cH cH
Secret Sharing Scheme and Key Sharing Scheme Suitable for Clustered Sensor Networks
179
7 CONCLUSIONS
We proposed three key sharing schemes. The user
can choose either of them according to the intended
usage.
REFERENCES
Heinzelman, W.R., Chandrakasan, A., Balakrishnan, H.,
2000. Energy-Efficient Communication Protocol for
Wireless Microsensor Networks, Proceedings of the
33rd Hawaii International Conference on System
Sciences, January 4-7.
Oliveira, L.B., Ferreira, A., Vilaça, M.A., Wong, H.C.,
Bern, M., Dahab, R., Loureiro, A.A., 2007.
SecLEACH—On the Security of Clustered Sensor
Networks, Signal Processing. Vol. 87, No. 12, pp.
2882-2895.
Qiang, T., Bingwen, W., Zhicheng, D., 2009. MS-LEACH:
A Routing Protocol Combining Multi-hop
Transmissions and Single-hop Transmissions, Pacific-
Asia Conference on Circuits, Communications and
Systems, pp. 107-110.
Bertier, M., Mostefaoui, A., Tédan, G., 2010. Low-Cost
Secret Sharing in Sensor Networks, Proceedings of the
IEEE 12th International Symposium on High
Assurance Systems Engineering (HASE'10), pp. 1–9,
November.
Yiying, Z., Chunying, W., Jinping, C., Xiangzhen, L.,
2013. A Secret Sharing-Based Key Management in
Hierarchical Wireless Sensor Network, International
Journal of Distributed Sensor Networks Volume 2013.
Kurihara, J., Kiyomoto, S., Fukushima, K., Tanaka, T.,
2008. On a Fast (k,n)-Threshold Secret Sharing
Scheme, IEICE Transactions on Fundamentals of
Electronics, Communications and Computer Sciences,
Vol. E91-A, No. 9, pp. 2365-2378, Sep.
Takahashi, S., Kang, H., Iwamura, K, 2014. Asymmetric
Secret Sharing Scheme Suitable for Cloud Systems,
2014 IEEE 11th Consumer Communications and
Networking Conference (CCNC), pp. 798-804Las
Vegas, Jan.
WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems
180