ENERGY-EFFICIENT SECURITY PROTOCOL FOR WIRELESS

SENSOR NETWORKS USING FREQUENCY HOPPING

AND PERMUTATION CIPHERING

Ismail Mansour, G´erard Chalhoub and Michel Misson

Clermont Universit´e/LIMOS CNRS, Complexe scientiﬁque des C´ezeaux, 63177 Aubi`ere cedex, France

Keywords:

Wireless sensor network, Security, Permutation ciphering, Frequency hopping, Energy efﬁciency.

Abstract:

The security aspect of wireless sensor networks has taken the attention of numerous researchers in the past

several years. It has recently been proven that public keys are now feasible in wireless sensor networks but

still consume a lot of processing time and memory. In this paper we propose the use of public keys based on

ECC to exchange symmetric keys that will be used to encrypt critical information. In addition, we propose a

time segmentation approach that enables the use of frequency hopping time slotted communications. Nodes

secretly exchange frequency hopping sequences that enable them to ﬁght against jamming and eavesdropping.

We use permutation ciphering technique to protect the information exchanged between nodes.

1 INTRODUCTION

Wireless sensor networks are more and more de-

ployed for various applications including home moni-

toring, health, industrial, military, etc. It is known that

wireless networks are easy to attack because of the na-

ture of the shared medium which makes it relatively

easy for intruders to eavesdrop, tamper or inject data

into the network. Sensor nodes are known to have

limited computation, storage, power and transmission

capacities, but attackers are not necessarily using the

same technology to launch their attacks.

The security level required might vary from one

application to another according to the importance of

the information that is being exchanged. In this paper,

we present a security mechanism using Elliptic Curve

Cryptography (ECC) to establish a secured communi-

cation based on frequency hopping in an energy efﬁ-

cient context. The main contribution of this paper is

the proposition of a secured protocolfor critical appli-

cations supporting a dynamic topology and hundreds

of nodes, and where security is a requirement and en-

ergy efﬁciency is a necessity. We present a crypto-

graphic system based on ECC that enables us to ex-

change secret frequency hopping sequences and en-

crypt exchanged data using a permutation ciphering

technique.

2 RELATED WORK

In this section, we go through the related work in the

ﬁeld of security in wireless sensor networks including

key management and frequency hopping.

2.1 Key Management in Wireless

Sensor Networks

In order to provide security in wireless sensor net-

works, communication should be encrypted and au-

thenticated. The main issue is how to set up secret

keys between nodes to be used for the cryptographic

operations, which is known as the key agreement.

Cryptographic systems are generally based on two

techniques: symmetric keys and asymmetric keys.

In symmetric key systems, nodes encrypt and de-

crypt messages using the same cryptographic key.

These systems are known for their light weight cryp-

tographic operations compared to public key systems,

thus they were the ﬁrst to be considered for wireless

sensor networks where computational resources are

scarce. One of the most known symmetric key sys-

tems that were proposed for wireless sensor networks

is SPINS (Perrig et al., 2002).

ZigBee is one of the most known standards for

wireless sensor networks. In the ZigBee-2007 spec-

iﬁcations (Zigbee, 2008), the protocol proposes a

277

Mansour I., Chalhoub G. and Misson M. (2011).

ENERGY-EFFICIENT SECURITY PROTOCOL FOR WIRELESS SENSOR NETWORKS USING FREQUENCY HOPPING AND PERMUTATION

CIPHERING .

In Proceedings of the 1st International Conference on Pervasive and Embedded Computing and Communication Systems, pages 277-282

DOI: 10.5220/0003371302770282

 SciTePress

cryptographic mechanism based on 128-bit symmet-

ric keys to secure communications between devices.

The most common key agreement for these systems is

key pre-distribution prior to network deployment. Af-

ter deployment, only the nodes sharing the same key

are able to communicate. Most of the key manage-

ment protocols for symmetric systems, lack resilience

to an observer that is able to listen to the discovery

process during which shared keys are exchanged.

Public key systems use a pair of keys: a private

key which is kept private and a public key which is

known by anyone. Data encrypted with one key is de-

crypted with the other, which avoids sharing the same

key for encryption and decryption. The most com-

mon criticism on using public key cryptography in

sensor networks is its computational complexity and

communication overhead. Nevertheless, recent work

such in (Hong, 2008) has been done to provethat pub-

lic key systems are feasible for wireless sensor net-

works where authors prove that public-key cryptog-

raphy based on ECC is very viable on small wireless

devices. ECC is based on the problem of logarithm

discrete. The main attraction of ECC over compet-

ing technologies such as RSA and DSA is the use of

smaller parameters but with equivalent level of secu-

rity (Lopez and Dahab, 2000). For example, 160-bit

ECC key is equivalent to 1024-bit RSA key. The per-

formance of ECC depends mainly on the efﬁciency of

ﬁnite ﬁeld computation and fast algorithms for elliptic

scalar multiplications.

2.2 Frequency Hopping as a Security

Mechanism

Frequency hopping is a known technique that en-

hances robustness against interferences in wireless

networks. By frequency hopping we mean the abil-

ity to change the frequency that is being used on the

physical layer from one transmission to another. Ben-

eﬁts of this technique in wireless sensor networks

have been proven in (Watteyne et al., 2009). Fre-

quency hopping can also be used to hide the commu-

nication against eavesdropping and this by making it

difﬁcult on an intruder to guess the frequency hopping

sequence.

In (Jones et al., 2003), the authors propose a holis-

tic approach to secure wireless sensor network by us-

ing symmetric key encryption and secured frequency

hopping sequences to protect and hide the commu-

nications. Frequency hopping sequences are updated

every now and then in such a way that an intruder has

not the time to guess the sequence. A sink node di-

vides the network into sectors and wedges, allocating

each sector a frequency hopping sequence and an en-

cryption key. In this paper, the applications covered

are only sink oriented data collection, and the network

size depends on the coverage area of the sink node.

Standardizations tendency (ISA, 2009; HART,

2008) towards frequency hopping is a proof that it

is an important in wireless communication in order

to make links more robust to interferences. It is also a

proof that the time synchronization needed for the slot

allocation is feasible as discussed in (Kerkez et al.,

2009). By adopting a secret frequency hopping se-

quence like in (Jones et al., 2003), this mechanism

can enhance security for wireless sensor network.

3 OUR PROPOSITION

A lot of attacks (Kavitha1 and Sridharan, 2010) like

sinkhole, black hole, spooﬁng and tampering can be

avoided using public key encryption infrastructure

that offers the means to ensure entity authentication,

source authentication, and data integrity and conﬁ-

dentiality. Our proposition is based on ECC encryp-

tion, frequency hopping and permutation ciphering.

We use ECAES encryption for ensuring data in-

tegrity and authentication, and we use session keys

(symmetric keys) for data encryption in addition to

permutation ciphering. Session keys and permuta-

tion ciphers are exchanged using ECAES encryption.

To ﬁght against jamming and interferences, we use

a frequency hopping mechanism based on shared se-

cret frequency hopping sequences. Each node in the

network negotiates, in a secured manner, a frequency

hopping sequence with all the nodes that it is allowed

to communicate with. This negotiation allows nodes

to choose the best frequencies as observed by the

nodes.

We consider a hierarchical topology as depicted

on ﬁgure 1, where the node S is the sink of the topol-

ogy and the default destination for the trafﬁc. We sup-

pose that the sink has more storage capacities than

other nodes. Hierarchical topologies are known to be

more convenient for energy efﬁciency in wireless sen-

sor networks (Akkaya and Younis, 2005). Nodes are

grouped into stars, each star has one central node that

we call router and several end-devices. Routers have

more processing and memory capacities then end-

devices and they constitute about 10 % of the network

size. End-devices represent the sensors and actuators

of the network. End-devices are only allowed to com-

municate with the router of the star to which they be-

long. The creation of stars is done during the net-

work deployment phase. Routers are allowed to com-

municate with each other but not with end-devices

that are not associated to them. This clustering tech-

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

278

nique is similar to the one used in IEEE 802.15.4 and

MaCARI (Chalhoub et al., 2008b). We suppose that

routers are able to aggregate data on the network layer

level.

Router

End-device

Figure 1: Network architecture. The network is decom-

posed in stars. Each star is under the control of one router.

Sensor and actors are associated to a router and belong to

only star.

3.1 Time Synchronization

for Frequency Hopping

and Energy Efﬁciency

As discussed earlier, frequency hopping helps against

interference and jamming attacks. In this paper, we

use a subset of the available frequencies to hop be-

tween them for two consecutive transmissions. In

order to establish a time synchronization that allows

nodes to hop from one frequency to another, we use

a synchronization mechanism based on beacon prop-

agation as used in the MaCARI protocol (Chalhoub

et al., 2008b; Chalhoub et al., 2008a).

Time is divided into cycles that start with a syn-

chronization period during which a beacon frame

containing synchronization information is propagated

along the topology by the routers. The propagation

starts at the sink and is done in a topological order in-

dicated inside the beacon to avoid collisions between

beacons. The sink allocates for each router a slot

in [T

] during which the beacon has to be propa-

gated to reach all the nodes of the network. This bea-

con contains time segmentation information allowing

nodes to know when to be active and when to save

energy.

Figure 2 depicts the time segmentation during a

cycle. In [T

], routers are allocated time intervals

during which they communicate with the end-devices

and then relays the collected data to the next router.

The time interval allocation is done during [T

]

by the sink according to the trafﬁc size generated in

each star. Once information is collected from end-

devices, routersexchange messages until the informa-

tion reaches its ﬁnal destination. Each router has to

keep track of the slots used for routing by neighbor

routers.

In order to allow new nodes to join the network in

a dynamic manner, we use CSMA/CA during [T

When a new node wants to join the network, it is not

yet known by the sink and does not have slots dur-

ing which it can communicate. This ensures a dy-

namic network where new nodes (cluster heads or

end-devices) are able to join even after the deploy-

ment phase. Nodes sends join request destined to the

sink during [T

]. Any router of the network is able

to receive the join request and relay it towards the sink

through other routers, this is explained in 3.2.

The cycle ends with an inactive period, [T

The duration of the inactive period depends on the

duty cycle of the nodes and the reactivity require-

ments of the application. This time segmentation al-

lows all the nodes of the network to save energy by

switching to sleep mode when they are not concerned

by the communication during a given time slot.

10 ms

Synchronization Period

CSMA/CA

Frequency Hopping Slots

Inactive

Period

Figure 2: Time segmentation for the frequency hopping

mechanism.

During the join phase (the phase during which

nodes are associated to the network), end-devices

and routers exchange the set of frequencies that they

would like to use for reception, they also inform the

sink of the trafﬁc they will generate at the application

layer level. The trafﬁc information enables the sink to

allocate enough time slots for the nodes. That is, ev-

ery entity is able to scan all possible frequencies and

decide which of these frequencies are less busy so it

is able to have a better reception rate using them. Fre-

quencies that seem good for the sender may not seem

good for the receiver.

At the end of the association phase, the router in-

dicates to the end-device the slots during which it will

communicate with it, the frequency to be used during

each slot and the communication direction (whether

it is a reception slot or a transmission slot). This ex-

change is encrypted using a session key created be-

tween the router and the end-device which is detailed

in 3.2. The duration of a communication slot is 10 ms,

which is enough time to send a frame of maximum

size and wait for its acknowledgment. This is the slot

duration of TSMP, WirelessHart and ISA100(Pister

and Lance, 2006; HART, 2008; ISA, 2009).

ENERGY-EFFICIENT SECURITY PROTOCOL FOR WIRELESS SENSOR NETWORKS USING FREQUENCY

HOPPING AND PERMUTATION CIPHERING

279

The order in which the frequency set is hopped is

speciﬁed by the router during the join phase. It is then

updated every l cycles, where l is maximum num-

ber of cycles tolerated before compromising the fre-

quency hopping order. So every l cycles, end-devices

receive a new order from the router encrypted using

the pairwise session key. l is a parameter that is spec-

iﬁed by the router during the join phase and can be

modiﬁed by the router when it updates the frequency

hopping order. During the cycle number l, the fre-

quency hopping slots are used in reception mode by

the end-devices.

3.2 Key Management

Our key management is based on three phases: before

deployment phase, join phase and neighbordiscovery.

The objective is to generate a symmetric key with ev-

ery neighbor for the routers, and a symmetric key for

every router and each of its end-devices.

3.2.1 Pre-deployment Phase

Keys are generated and predistributed before deploy-

ment. Nodes in a wireless sensor network come

from the same manufacturer, unlike the internet where

nodes are very heterogeneous. This feature makes it

easier to be able to distribute keys to nodes in a con-

trolled manner. For each node in the network (routers

and end-devices) we generate a pair of asymmetric

keys (a public key and a private key). We install these

two keys in addition to the public key of the sink in

each node of the network.

3.2.2 Join Phase

First step after the deployment is joining phase. Let

us consider a new router R that wants to join the net-

work through the router F (see ﬁgure 1). R sends a

join request message (joinReq) to F containing fre-

quency hopping information (the set of frequencies

that R wants to use for reception) and its public key

). This join request message is encrypted with

the public key of S (P

) using ECAES as described

in (Lopez and Dahab, 2000). F receives the join re-

quest message and relays it to C in order to reach the

sink node S. S examines the join request and decide

whether to accept it or not. In case it is accepted, S

sends back a join response encrypted using the pri-

vate key of S (K

). When F receives the join response

encrypted with the private of S, it can be sure that

the sink has accepted that R joins the network. F de-

crypts the join response using the public key of S (P

)

and retrieves the public key of R. Then, F sends to

R the join response encrypted with the public key of

R. This join response includes the public key of F

) and a session key (SK

). The session key SK

is used to encrypt communication between F and R.

This communication is detailed on ﬁgure 3.

R C

( joinReq)

( joinRes)

(dataR)

(dataF)

Figure 3: Join phase.

3.2.3 Neighbor Discovery

The sink plays a critical role in the join phase. It

needs to authenticate the public keys of every node

that wants to join the network. When this authentica-

tion fails, it sends back a reject message to that node.

Otherwise, it sends a join response. The sink saves

all the public keys of the nodes in the network. This

way, whenever a node wants to know the public key

of another node, it sends a public key request to the

sink with the identiﬁer of the other node (the identi-

ﬁer can be its physical address). The sink sends back

the public key encrypted using its private key.

Session key exchange is done with every neighbor

detected by the routing protocol. In the case of session

key creation between neighboring routers (for exam-

ple, routers A and B in ﬁgure 1), the node that creates

the session key is the node with the higher priority.

The priority of a node is derived from its identiﬁer.

The node with the smaller identiﬁer has the higher pri-

ority. When node A identiﬁes node B as a neighbor,

it checks if it has a higher priority. If it is the case,

A sends a public key request (pkReq) to the sink to

obtain the public of node B (P

). S sends back a pub-

lic key response (pkRes) to A containing the public

key of B (P

). Then, A sends a session key establish-

ment (skEst) message to B encrypted with P

includ-

ing the frequency hopping information, a session key

(SK

B) and the public key of A (P

). B replies with an

acknowledgment (ACK) including its frequency hop-

ping information. Frequency hopping information in-

dicates the time slots used by a router to communicate

with other routers, the frequency and the permutation

cipher used in each slot. If B has the higher priority,

it is the other way around. This phase is represented

on ﬁgure 4.

Session keys are then updated every n cycles,

where n is the maximum number of cycles tolerated

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

280

A S

(pkReq)

(pkRes)

(skEst)

(ACK)

(dataB)

(dataA)

Figure 4: Neighbor discovery phase.

before compromising the session key. In order to

make encrypted diffusion possible using symmetric

cryptography, the sink creates a global session key.

This session key is then propagated to all the routers

of the network in a hop by hop manner along the par-

ent child relationship starting from the sink. This ses-

sion key is used do encrypt diffused messages such

as the beacons of the synchronization period. The

global session key is updated every k cycle and dif-

fused in the beacon frame, where k is the maximum

number of cycles tolerated before the global session

key is compromised. The new global session key is

encrypted using the old session key. Local common

session keys could also be created for each star by the

router of the star. This local session key is used for

diffusion inside the star. Each router speciﬁes a re-

served slot that is used for diffusion inside the star on

a certain frequency.

3.3 Permutation Ciphering

As cryptographic operations take a lot of time on sen-

sor nodes, we only encrypt critical information using

the session keys. To accelerate the data exchange,

we use permutation ciphering. Indeed, the frequency

hopping information contains, in addition to the fre-

quency used in each slot, a permutation cipher that

enables nodes to encrypt data that they exchange. We

suppose that the permutation should change from one

slot to another. So for each slot a node should specify

a permutation cipher and a frequency.

The permutation cipher is applied on words. Each

word is a group of 31 symbols (a symbol is a group

of 4 bits). The rationale behind this grouping is that

the IEEE 802.15.4 physical layer uses the O-QPSK

modulation were data is grouped in symbols (IEEE

802.15, 2006).

When a node has a data to transmit (whether its

an end-device or a router), it divides it into 8 words

composed of 31 symbols each as explained on ﬁgure

5. In this manner, we are able to apply a permutation

maximum frame size of 124 bytes of data on the MAC

level. When the data that a node needs to send is less

than 124 bytes, the frame is stuffed with bits of 0.

S0 S1 S2 S3 S4 S5 S6 S7

...

S243

S240

S242

S244 S245

S5S6

S246

S247

S11

S241

S240 S247

...

S246S245S244S243S242S241

Word 1 Word 3 Word 5 Word 7

31 lines

Figure 5: Word permutation using the permutation cipher

(2, 7, 5, 0, 1, 4, 3, 6).

The transmitted data will be the concatenation of

the words. Hence, in the example shown on ﬁgure

5, the order in which the words are transmitted is the

following: word 3, word 4, word 0, word 6, word 5,

word 2, word 7, word 1.

In order to decipher the received frame, a node

needs to have the permutation cipher. The permu-

tation cipher is deduced from the frequency hopping

information exchanged either during the join phase or

the neighbor discovery phase. This information con-

tains the frequency used during each slot and the per-

mutation cipher associated to that slot.

4 SECURITY AND ENERGY

EFFICIENCY ANALYSIS

In what follows we go through the different security

aspects that are taken into considerations in this archi-

tecture.

Conﬁdentiality. During the join phase after the de-

ployment, exchanges are encrypted using the ECAES

protocol. Once pairwise session keys are established

between nodes, data is exchanged and encrypted us-

ing a symmetric key using AES encryption for critical

data like frequency hopping information and beacons,

and application data is encrypted using the permuta-

tion cipher.

Source Authentication and Integrity. Message in-

tegrity and source authentication can be ensured using

ENERGY-EFFICIENT SECURITY PROTOCOL FOR WIRELESS SENSOR NETWORKS USING FREQUENCY

HOPPING AND PERMUTATION CIPHERING

281

ECDSA for generating the hash of the message and

signing it using the private key of the sender.

Availability. In our solution the intruder can only

paralyze the system by physically accessing the in-

dividual sensor nodes. Whereas our frequency hop-

ping mechanism is resilient against a denial of ser-

vice attack. It makes it hard on an attacker to jam

the network, it will need to jam over all the available

channels simultaneously. Even if it does so, when

the MAC layer is not able to ﬁnd a free channel to

use, it will stop sending messages. By not sending

messages, the network supervisor is able to notice the

malfunction and suspect a jamming over all the avail-

able channels.

Energy-efﬁciency. The time segmentation ap-

proach allows nodes to save energy and sleep during

the time slots where they are not concerned by the

communication. End-devices are active only during

the time slots speciﬁed by the router to which they are

associated. A router is active during the time slots of

its end-devices, the time slots of its router neighbors

and [T

]. In addition, all nodes save energy during

5 CONCLUSIONS

In this paper we presented a energy-efﬁcient security

management for wireless sensor networks based on

pre-distributing ECC public keys before deployment.

These public keys are then used to create session keys

to encrypt the exchanged critical information such as

frequency hopping information and permutation ci-

phers. In addition, we proposed to use frequency

hopping in a secure way by exchanging secret fre-

quency hopping sequences which enable nodes to re-

sist against deny of service attack and eavesdropping.

Since cryptographic operations are very expensive in

wireless sensor networks, we propose a permutation

ciphering technique to protect data.

Our solution is dynamic as it allows nodes to join

the network at any time after deployment given that

this node has a valid public key. In our proposition,

the sink node does not have to be in range of the

routers. The sink node can reach all the routers of

the network in a guarantees hop-by-hop manner. The

security aspect of our proposition is still in progress.

We are working on evaluations on telosB motes.

This work is partially funded by FEDER (Euro-

pean fund for regional development).

REFERENCES

Akkaya, K. and Younis, M. (2005). A survey on routing

protocols for wireless sensor networks. Ad hoc net-

works, 3:325–349.

Chalhoub, G., Guitton, A., Jacquet, F., Freitas, A., and Mis-

son, M. (2008a). Medium access control for a tree-

based wireless sensor network: Synchronization man-

agement. In IFIP Wireless Days.

Chalhoub, G., Guitton, A., and Misson, M. (2008b). MAC

speciﬁcations for a WPAN allowing both energy sav-

ing and guaranted delay - Part A: MaCARI: a synchro-

nized tree-based mac protocol. In IFIP WSAN.

HART (2008). HART ﬁeld communication protocol spec-

iﬁcations. Technical report, HART Communication

Foundation Std.

Hong, P. (2008). Feasibility of pkc in resource-constrained

wireless sensor networks. In ICCIT.

IEEE 802.15 (2006). Part 15.4: Wireless medium ac-

cess control (MAC) and physical layer (PHY) spec-

iﬁcations for low-rate wireless personal area networks

(WPANs). Standard 802.15.4 R2006, ANSI/IEEE.

ISA (2009). ISA100.11a:2009 wireless systems for indus-

trial automation: Process control and related applica-

tions. Draft standard, in preparation, International So-

ciety of Automation Std.

Jones, K., Wadaa, A., Oladu, S., Wilson, L., and Eltoweissy,

M. (2003). Towards a new paradigm for securing

wireless sensor networks. In New Security Paradigms

Workshop.

Kavitha1, T. and Sridharan, D. (2010). Security vulnerabil-

ities in wireless sensor networks: A survey. Journal

of Information Assurance and Security, 5:31–44.

Kerkez, B., Watteyne, T., Magliocco, M., Glaser, S., and

Pister, K. (2009). Feasibility analysis of optimal con-

troller design for adaptive channel hopping. In WS-

NPERF.

Lopez, J. and Dahab, R. (2000). An overview of ellip-

tic curve crytography. Technical report, University of

Campinas.

Perrig, A., Szewczyk, R., Tygar, J., Wen, V., and Culler,

D. (2002). SPINS: Security protocols for sensor net-

works. Wireless Networks.

Pister, K. and Lance, D. (2006). TSMP: time synchronized

mesh protocol. In Distributed sensor networks.

Watteyne, T., Mehta, A., and Pister, K. (2009). Reliabil-

ity through frequency diversity: Why channel hopping

makes sense. In PE-WASUN.

Zigbee (2008). Zigbee Speciﬁcation. Zigbee Standard

053474r17, ZigBee Standards Organization.

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

282