SECURE AD-HOC ROUTING THROUGH A-CODES
Giovanni Schmid
1
and Francesco Rossi
2
1
High Performance Computing and Networking Institute (ICAR), Naples, Italy
2
University of Naples Parthenope, Naples, Italy
Keywords:
Message authentication, Secure routing, Ad-hoc wireless networks.
Abstract:
Wireless ad-hoc networks are very attractive in several application domains, but the very nature of these
networks and their cost objectives pose big security challenges, perhaps making them the most difficult net-
working environments to secure. A particular challenging issue is that of secure routing. In this work we
propose to get secure routing for such networks through a special coding technique at the physical layer of
radio communication channels. This approach has the main advantage of being applicable to any routing pro-
tocol, without requiring modifications to the protocol itself, but with a suitable key management. We illustrate
it for the concrete case of AODV, the standard routing protocol for Low-data-Rate Wireless Personal Area
Networks (LR-WPANs). The resulting analysis seems to indicate that such approach is very promising for
LR-WPANs, both in term of performance and energy efciency.
1 INTRODUCTION
Wireless ad-hoc networks are very attractive for many
applications, but the absence of a fixed infrastructure,
the adaptivity and time-varying shape of their inter-
connection meshes, together with the physical char-
acteristics of the radio channel, combine to create sig-
nificant challenges, perhaps making them the most
difficult networking environments to secure. In addi-
tion, node deviceshave limited capabilities in terms of
computing power, available storage, and power drain,
putting severe limits on the security overhead these
networks can tolerate, something that is of far less
concern with higher bandwidth networks.
A particular challenging issue for wireless ad-hoc
networks is that of designing both secure and effi-
cient routing protocols. The reason for that is twofold.
First, there are a variety of ad-hoc wireless networks
(e.g. sensor networks, vehicular networks) and usage
scenarios (e.g. ambient monitoring, industry control),
so it is impossible to provide a single routing proto-
col fitting each need. Last but not least, the design of
reliable and efficient secure routing protocols for real-
word applications has resulted to be an elusive issue.
These difficulties are undoubtedly reflected by
current emerging industry standards. The ZigBee Al-
liance
1
simply adopts the Ad hoc On Demand Distan-
1
An association of companies founded in 2003 and com-
ce Vector (AODV) routing protocol (Perkins et al.,
2003), which does not provide any authentication
mechanism at all, and is vulnerable to many attacks
(Sanzgiri et al., 2002; Ning and Sun, 2003).
In this work, we propose to secure routing in wire-
less ad-hoc networks through A-coding. The A-code
primitivewas introduced in (Schmid and Rossi, 2010)
to allow for the establishment of authentic public keys
in wireless sensor networks. Here, we exploit A-
coding directly at the physical layer of the radio com-
munication channel, as a low level message authen-
tication processing. This approach turns out in offer-
ing hop-by-hop message authentication to any routing
protocol without requiring modification to the proto-
col itself, but just providing appropriate key manage-
ment at the network layer.
We illustrate the A-coding approach for the con-
crete case of AODV, sketching a comparison with
the coding techniques of the IEEE Standard 802.15.4
(IEEE, 2006), which has been adopted by ZigBee for
the specifications of the medium access control sub-
layer (MAC) and the physical layer (PHY) of the LR-
WPAN protocol stack. The resulting analysis seems
to indicate that such approach is very promising for
LR-WPANs, both in term of performance and energy
efficiency.
The rest of the paper is organized as follows. Sec-
mitted in defining the protocol stack for low-data-rate, wire-
less networks (LR-WPANs); http://www.zigbee.org
151
Schmid G. and Rossi F..
SECURE AD-HOC ROUTING THROUGH A-CODES.
DOI: 10.5220/0003313701510156
In Proceedings of the 1st International Conference on Pervasive and Embedded Computing and Communication Systems (PECCS-2011), pages
151-156
ISBN: 978-989-8425-48-5
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
tion 2 discusses about related work, whilst Section 3
describes the A-code primitive. In Section 4 we show
how to secure AODV thanks to A-coding. Section 5
sketch a comparison with the 802.15.4, and discusses
general A-code key management issues for AODV.
Finally, in Section 6 we draw out our conclusions, and
illustrate our future work.
2 RELATED WORK
Attacks against routing can be realized through a
suitable combination of offensive techniques such
as eavesdropping, identity spoofing, and the replay,
modification, forgery or deletion of routing control
packets. The adaptive nature of the communica-
tion mesh in ad-hoc networks, along with node’s
constrained resources, compel the adoption of on-
demand routing, which in case of wireless communi-
cations is particularly exposed to such offensive tech-
niques.
Many secure routing protocols for wireless ad-
hoc networks, based on both private-key and public-
key cryptography, have been proposed by the research
community over the last decade, with alternate re-
sults as in (Sanzgiri et al., 2002; Zapata and Asokan,
2002; Papadimitratos and Hass, 2003). Reporting on
such proposals is outside the scope of this work. A
comprehensive survey of secure on-demand routing
is given in (Hu and Perrig, 2004), whilst (Karlof and
Wagner, 2003) describes attacks on sensor network
routing protocols, and introduces some generic coun-
termeasures.
In virtually all cases, strategies have been adopted
to face against node’s constrained resources and
achieve acceptable overheads. Those approaches,
not surprisingly, have often resulted in some secu-
rity weakness or in assumptions that are difficult to
be satisfied in practice (Ramachandran and Yasinsac,
2001). For example, in (Sanzgiri et al., 2002; Hu
et al., 2003; Hu et al., 2005) authentication is real-
ized only for routing control packets, exposing data
packets to serious threats. Both SAODV (Zapata and
Asokan, 2002) and ARAN (Sanzgiri et al., 2002) pro-
vide message authenticity only when all intermediate
nodes are trustworthy, which is an overly restrictive
assumption. The TESLA authentication framework
(Perrig et al., 2000) - along with its variant µTESLA
(Perrig et al., 2002), specifically designed for wire-
less sensor networks, avoid hop-by-hop authentica-
tion by relying on loosely time synchronized network
nodes. However, secure time synchronization has
been demonstrated to be very difficult to achieve, also
on networks with a fixed infrastructure (Menezes, et
al.,1996). Consequently, all the routing protocols
based on such frameworks(e.g. the protocol in SPINS
(Perrig et al., 2002), SEAD (Hu et al., 2003), SEAR
(Zhao et al., 2008)) suffer the same drawback.
Considered together, the above works and expe-
riences seem to indicate that effective secure proto-
cols for ad-hoc routing can only be achieved if hop-
by-hop authentication is guaranteed for all the pack-
ets involved in the protocol. Both LHAP (Zhu et al.,
2003) and HEAP (Akbani et al., 2008) were designed
to offer hop-by-hop authentication for data packet as
well as control packets. LHAP realizes message au-
thentication through one-way hash key chains, and
that turns out in a low efficiency in terms of mem-
ory requirements, since long time communications re-
quires long chains. HEAP is a modified version of the
HMAC algorithm that uses two keys and seems very
efficient for multicast communications. HEAP was
designed to defend against attacks originating from
nodes that are not authenticated members of the net-
work (outsider attacks).
The A-coding approach offers hop-by-hop au-
thentication, too, and in the same security assump-
tions than HEAP. However, it has one main advan-
tage: since hop-by-hop authentication is realized at
the lowest layer of the protocol stack, any protocol
can get its own proper authentication service by just
being coupled with suitable key management, without
any modification to the protocol itself.
A-codes are based upon I-codes (Cagalj et al.,
2006) and, like these ones, were introduced to allow
for the establishment of authentic public keys over in-
secure radio channels (Schmid and Rossi, 2010). Dif-
ferently than I-codes, however, A-codes can provide
authentication without user intervention and in the ab-
sence of special, dedicated radio-frequency channels
(integrity channels).
3 THE A-CODE PRIMITIVE
The A-code primitive was introduced in (Schmid and
Rossi, 2010) to allow for the establishment of authen-
tic public keys in wireless sensor networks. It can
operate directly at the physical layer of the protocol
stack, on PPDUs
2
, offering a “physical coding” al-
ternative to traditional message authentication codes
(MACs).
In A-codes, message integrity is gained through
unidirectional message coding and on-off keying
communication with signal anti-blocking; these are
2
PPDU stands for PHY Protocol Data Unit, and repre-
sent the message structure managed by the IEEE Standard
802.15.4 (IEEE, 2006) physical (PHY) layer.
PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems
152
the three components which give rise to Integrity
Codes (Cagalj et al., 2006).
Unidirectional message coding ensures that bit 0
cannot be changed in bit 1. Manchester code is an
example of unidirectional coding scheme; it encodes
each bit 1 as 10 and each bit 0 as 01. If we suppose
that an adversary can only convert a bit 0 into bit 1,
then the receiver will be able to detect forged mes-
sages, since such messages cannot be decoded prop-
erly.
On-off keying (OOK) is a signal modulation tech-
nique such that bit 1 is transmitted as the presence
of a signal, and bit 0 as the absence of a signal for a
known time slice. Signal anti-blocking uses a random
energy signal, so that an adversary cannot annihilate
bits 1 by jamming the signal. Considered together,
on-off keying and signal anti-blocking give a good re-
sistance against attacks based upon jamming and/or
bit-flipping.
I-Codes can also achieve message authentication,
but only through presence awareness. That is, the
receiver needs to be aware of the fact that the re-
ceived signal is on the channel used by an authorized
sender. This requires an infrastructure of authorized
senders located in known positions or, alternatively,
continuously signaling senders on known channels.
Of course, these two conditions are not usually sat-
isfied in case of LR-WPANs.
In A-codes, authentication for each single PPDU
is provided by expanding some of its bits into pseudo-
random strings p(n) given by:
p(n) = H(KkH(KkH(....H(KkX
ID
)....)))
= H(Kkp(n 1)) . (1)
In (1), X
ID
denotes the identifier of node X, H is a
suitable and known cryptographic hash function, n is
an integer related is some way to the number of ex-
panded bits sent so far by node X to the receiver node
Y, and K is a secret key shared between X and Y. The
function H is applied iteratively n times. The receiver
Y will consider a given PPDU authentic only if:
H(Kkp
(n 1)) = p(n) , (2)
for any of the values of n corresponding to one of the
expanded bits in such PPDU. In equation (2) p
rep-
resents the pseudo-random string as computed by Y,
whereas - because of the message integrity property
assured by OOK with unidirectional coding and sig-
nal antiblocking - p(n) is the pseudo-random string
issued by X. The first member of (2) can be com-
puted by Y as a consequence of its knowledge of H,
K, X
ID
and n. Since the value of n has not to be kept
secret, the synchronization between X and Y can be
easily obtained by sending that value in clear from X
to Y in each PPDU. For efficiency and easy of im-
plementation, only a constant, small number of bits
at prescribed off-set positions in each PPDU should
be expanded through (1). Actually, because of the in-
tegrity property of PPDUs, just one single bit could
be expanded for each PPDU; however, noisy chan-
nels could require a slightly greater number of hash
expansions to avoid denial of service conditions.
4 SECURING AODV
In this section we illustrate how to secure the AODV
routing protocol with the A-Code primitive. Our ar-
guments could have been easily exposed for the gen-
eral case of (on-demand) routing protocols, without
any substantial modification. However, AODV is one
of the most prominent protocols for ad-hoc wireless
networks, and discussing it explicitly will allow us
to be more specific in the description of our tech-
nique and to sketch a comparison, in terms of bit rate
throughput, with the modulation techniques provided
for the ZigBee protocol stack.
AODV is based on three standard routing mes-
sages (see Figure 1): route request (RREQ), route
replay (RREP) and route error (RERR). The RREQ
packet is used when a route to a new destination is
needed: the source node broadcasts an RREQ packet
to find a route to the destination. The route is made
available by unicasting a RREP packet back to the
originator of the RREQ. Each node receiving the re-
quest caches a route back to the originator of the re-
quest, so that the RREP can be unicast from the desti-
nation along a path to that originator, or likewise from
any intermediate node that is able to satisfy the re-
quest. A RERR message is used when a link break
in an active route is detected, in order to notify other
nodes that some destinations are no longer reachable
by way of the broken link. The format lengths for
RREQ, RREP and RERR messages in AODV are of
24, 20 and 20 bytes, respectively.
We limit our description of A-coding to RREQ
packets, since the other two cases are very similar.
The A-coding workflow composes of the following
steps (see Figure 2):
1. The first step consists in encapsulating in a source
PPDU the MAC data frame (MPDU) contain-
ing the RREQ packet as its own MAC payload
field. We remark that during this phase nor the
RREQ packet is modified, neither specific fields
are required in the MPDU. The output is a PPDU
which has the same structure and similar infor-
mation encoding as prescribed by IEEE Standard
802.15.4, but has a slightly modified synchro-
SECURE AD-HOC ROUTING THROUGH A-CODES
153
Figure 1: Route Request (RREQ), Route Replay (RREP)
and Route Error (RRER) message formats for AODV.
nization header SHR’, which replaces the IEEE
SHR for transceiver synchronization in A-coding
mode;
2. According to the information encoded in SHR’, m
bits (the figure shows the case m = 3) at prescribed
offset positions in the source PPDU are expanded
into hash digests of size |H| given by (1). The out-
put of this phase is an “expanded PPDU which is
m|H| bytes longer than the source PPDU;
3. After the expansion through hash digests, the
PPDU is coded thanks to an unidirectional cod-
ing scheme. For Manchester scheme the rate is
r = 2, but schemes exist such that r 1.5 ;
4. Finally, the bits constituting the output chip
stream from the previous phase are modulated by
using OOK with signal antiblocking. The length
of the chip stream is given by r(|SHR
|+ |PHR|+
|PSDU| +m|H|), where the three first addendums
are the lengths of the (modified) synchronization
header, the PHY header and the PHY payload, re-
spectively.
5 PERFORMANCE AND
SECURITY ISSUES
In order to show the viability of our approach, this
section discusses about performance and security
goals of A-codes. Section 5.1 gives a performance
comparison of A-coding versus the related techniques
encompassed by IEEE Standard 802.15.4, which - as
Figure 2: A-Coding workflow for the RREQ packet of
AODV.
we told in the introduction - defines the medium ac-
cess sublayer (MAC) and physical layer (PHY) spec-
ifications for the LR-WPAN protocol stack. Different
alternatives exist in setting up cryptographic keys in
view of node authentication, which affect the depend-
ability of the network. This issue is briefly discussed
in Section 5.2 for the case of securing AODV through
A-codes.
5.1 A-coding versus IEEE 802.15.4
IEEE PHY layer (IEEE, 2006) uses spread-spectrum
modes of communication to obtain some resilience to
jamming and improved noise immunity. In alterna-
tive to spread spectrum, A-codes use the spreading
technique illustrated in Section 4, so in both cases
it is possible to consider the spread factor (SF), that
is the ratio of the chipstream length to that of its re-
lated source PPDU. The spread factor is a measure
of loss in data throughputs w.r.t. the rates at which a
transceiver can receive and transmit chips. Thus, the
gain factor
γ =
SF
I
SF
A
can be used to compare the performance, in term of
data rates, of a given IEEE coding technique (I) ver-
sus the A-coding one (A), assuming that a transceiver
can perform at roughly the same chipstream rates for
I and A. A value γ < 1 indicates that the considered
IEEE technique outperforms A-coding, and vicev-
ersa. Figure 3 reports the value of γ relative to the
PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems
154
RREQ packet of AODV for all the coding techniques
at the PHY layer (IEEE, 2006). Those values were
computed for a unidirectional code of rate r = 1.5 and
for a number of m = 1, 2, 3 bits subjected to an hash
digest expansion of 128 bits. For the figures, we as-
sumed for simplicity |SHR| = |SHR
|, and |MHR| = 7
(see Section 4).
Figure 3: Values of the gain factor γ of A-coding w.r.t.
the coding techniques encompassed by the IEEE Standard
800.15.4 PHY layer.
Figure 3 shows that in many cases A-coding out-
performs spread-spectrum modes of coding. When
that does not occur (as in the case of 868 ASK, be-
cause of its very low spreading factor of 1.6), the
A-coding performance, however, widely falls in the
admissible data-ranges for LR-WPANS. Indeed, the
data rate achieved in the worst case (m = 3) for an
OOK modulation performed at only 300 Kchip/s is of
32.4 Kb/s, an improvement of more than 60% on the
lower bound (20 Kb/s) prescribed by the standard.
On the other hand, OOK transceivers are nowa-
days on the market that support data rates of hun-
dreds of Kb/s. These transceivers are cheap and have
very low power consumptions, because of the sim-
plicity of OOK implementation and the fact that de-
vices can save on transmit power when sending 0’s
as no signal. As a matter of fact, OOK modula-
tion is emerging as the optimal choice in short-range,
battery-operated wireless applications such as wire-
less sensor networks.
5.2 Key Management
Authentication through A-codes relies ultimately
upon the secret key K (see equation 1), and the way
K is established, shared and managed by the network
nodes relates to the threat model assumed for that net-
work, resulting in different levels of security. As a
general matter of fact, it is unpractical to achieve se-
curity in an absolute sense (if any); rather, the point
is to get a good trade-off between protection, usabil-
ity and performance. Discussing the many proposals
and contributions given in the literature on the issue
of (secret) key set up and management is by far out
the scope of this work. We sketch here just few con-
siderations for the specific case of AODV, under the
underlying assumption that only outsider attacks are
possible
3
.
A very simple possibility is to use the same key
K for all nodes in the network. In this case K plays
the role of a network key, and key management can
be very lightweight by loading a first key instance
on each node during the deployment stage, and then
updating it at scheduled times via broadcast commu-
nication. With this scheme, thanks to A-codes each
node in the network has corroborate evidence that a
receiving packet is fresh and was generated by an-
other node in the network
4
. Moreover, this scheme
implicitly supports multicast communications, which
is the case for any RREQ request and any RERR mes-
sage.
Against the above advantages, this scheme suf-
fers however the following three main drawbacks: (a)
breaking the key may turn out in the compromission
of the entire network; (b) detecting and locating unau-
thorized routing paths and functions can be very dif-
ficult, and; (c) it is very expensive to recover from a
compromission.
Thus, sensitive usage contexts and exposed envi-
ronments require a more complex (and resource con-
suming) key management schema, with one or more
keys for each node. A standard approach is that of
using different kinds of key, each one with its specific
protection role and its own establishment and man-
agement protocols (Menezes et al., 1996). In all these
cases, the key K in expression (1) should play the role
of a link key in case of unicast communications (as for
RREP packets), and of a group keyin case of multicast
communications (i.e., RREQ and RERR packets). In
the first case, an instance for K will be generated and
shared only among two one-hop-neighbouring nodes,
with the scope of authenticating the communication
link between them. In the second case, instead, K will
be shared among a given node X and all its one-hop
neighbours, in order to authenticate the packets from
X.
3
By the way, the best one can do in facing insider at-
tacks, that is attacks by nodes running malicious code or
adversaries who have stoled the key material, is to provide
for intrusion detection and failure detection.
4
The parameter n in expression (1) plays a crucial role
here, since it assures that unauthorized nodes cannot per-
form packet relying and replaying.
SECURE AD-HOC ROUTING THROUGH A-CODES
155
6 CONCLUSIONS AND FUTURE
WORK
In this work we propose to secure routing in wireless
ad-hoc networks through A-coding, a special coding
technique which operates directly at the physical layer
of the radio communication channel. A-coding acts
as a low level message authentication processing, and
constitutes an alternative to traditional message au-
thentication codes. A performance comparison of A-
coding versus the related coding techniques encom-
passed by IEEE 802.15.4, the adopted standard for
low-data-rate, wireless networks, shows the effective-
ness of this approach. On the other hand, A-coding
relies on OOK modulation, and OOK transceivers are
nowadays on the market that have very low costs and
power consumptions.
Our future work will be focused on a prototypal
implementation of A-coding as firmware in a com-
mercial off-the-shelf transceiver, in order to carry out
a testbed for the adoption of this technology.
REFERENCES
Akbani, R., Korkmaz, T., and Raju, G. (2008). HEAP: A
packet authentication scheme for mobile ad-hoc net-
works. Ad Hoc Networks, 6(7).
Cagalj, M., Hubaux, J., Capkun, S., Rengaswamy, R.,
Tsigkogiannis, I., and Srivastava, M. (2006). Integrity
(i) codes: Message integrity protection and authen-
tication over insecure channels. In Proceedings of
the 2006 IEEE Symphosium on Security and Privacy.
IEEE.
Hu, Y., Johnson, D., and Perrig, A. (2003). SEAD: Secure
efficient distance vector routing for mobile wireless ad
hoc networks. Ad Hoc Networks, 1(1).
Hu, Y. and Perrig, A. (2004). A survey of secure wireless
ad hoc routing. Security & Privacy, 2(3).
Hu, Y., Perrig, A., and Johnson, D. (2005). Ariadne: A se-
cure on-demand routing protocol for ad hoc networks.
Wireless Networks, 11(1-2).
IEEE (2006). Std IEEE 802.15.4 - Part 15.4: Wireless
Medium Access Control (MAC) and Physical Layer
(PHY) Specifications for Low-Rate Wireless Personal
Area Networks (WPANs).
Karlof, C. and Wagner, D. (2003). Secure routing in wire-
less sensor networks: attacks and countermeasures.
Ad Hoc Networks, 1(1-2).
Menezes, A., van Oorschot, P., and Vanstone, S. (1996).
Hanbook of Applied Cryptography. CRC Press.
Ning, P. and Sun, K. (2003). How to misuse AODV: A case
study of insider attacks against mobile ad-hoc routing
protocols. In Information Assurance Workshop 2003.
IEEE.
Papadimitratos, P. and Hass, Z. (2003). Secure routing for
mobile ad hoc networks. Mobile Computing and Com-
munications Review, 7(1).
Perkins, C., Belding-Royer, E., and Das, S. (2003). RFC
3561 - Ad Hoc On Demand Distance Vector (AODV)
Routing. IETF.
Perrig, A., Canetti, R., Tygar, D., and Song, D. (2000). Ef-
ficient authentication and signing of multicast streams
over lossy channels. In Proceedings of the IEEE Sym-
posium on Research in Security and Privacy. IEEE.
Perrig, A., Szewcyzk, R., Tygar, J., Wen, V., and Culler,
D. (2002). SPINS: security protocols for sensor net-
works. Wireless Networks, 8(5).
Ramachandran, P. and Yasinsac, A. (2001). Limitations of
on demand secure routing protocols. In Workshop on
Information Assurance. IEEE.
Sanzgiri, K., Dahill, B., Levine, B., Shields, C., and
Belding-Royer, E. (2002). A secure routing proto-
col for ad hoc networks. In Proceedings of the 10th
IEEE International Conference on Network Protocols.
IEEE.
Schmid, G. and Rossi, F. (2010). A-code: A new crypto
primitive for securing wireless sensor networks. In
Proceedings of the First International Workshop on
Trust management in P2P Systems. Springer.
Zapata, G. and Asokan, N. (2002). Securing ad hoc routing
protocols. In Proceedings of the 1st ACM workshop
on Wireless security. ACM.
Zhao, M., Walker, J., Hu, Y., Perrig, A., and Trappe, W.
(2008). SEAR: A secure efficient ad hoc on demand
routing protocol for wireless networks. In Proceed-
ings of ASIACSS ’08. ACM.
Zhu, S., Xu, S., Setia, S., and Jajodia, S. (2003). LHAP:
a light-weight hop-by-hop authentication protocol for
ad-hoc networks. In Proceedings of the 23rd ICDCS
Workshop. IEEE.
PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems
156