![](bg3.png)
2.425 2.4252 2.4254 2.4256 2.4258 2.426 2.4262
0
0.2
0.4
0.6
0.8
1
Jamming Center Frequency (GHz)
Packet Delivery Ratio
One Filter
No Filter
Figure 3: We show the packet delivery ratio for a vanilla
receiver and a receiver with a jamming mitigation filter at
the base-band. The use of only a high-pass filter defends
against attacks at the channels center frequency, while the
filter is detrimental to the receiver when attacks are centered
at other frequencies.
The 802.15.4 protocol uses a 2 byte cyclic redun-
dancy check to test for packet errors (IEEE 802.15.4,
2006). This is calculated at the transmitter and ap-
pended to the packet. The receiver raises a flag if there
is an error and passes the data up the stack if no error
is detected. The use of the 2 byte CRC can pose a vul-
nerability since an intelligent attacker can completely
jam an 802.15.4 system using the two byte CRC by
only jamming one symbol in every packet.
We consider a periodic jamming attack which al-
ternates between a sleeping and attacking state. This
attack is advantageous because it looks to take advan-
tage of the weakness introduced in using a two-byte
CRC. The periodic jamming attack also reduces en-
ergy to allow for attacks to be mounted longer from
low-power devices. This attack is often assumed to
occur at the center frequency of the legitimate chan-
nel. We do not make this assumption but rather al-
low the attacker to choose any center frequency and
to occasionally change its center frequency within the
channel.
We revisit the model which is used to motivates
using a digital filter to mitigate a baseband jamming
attack (DeBruhl and Tague, 2011). Using this model
they arrive at the plot shown in Figure 1, which moti-
vates the fact that a filter can be used to mitigate the
effect of periodic jamming at the base band.
To further motivate this fact, we present Fig-
ure 2, showing the cumulative normalized power for
802.15.4 and a periodic jammer in for a bandwidth
centered around the carrier frequency. This plot
shows 85% of of the attacker’s power is contained in a
region with only 10% of the signal power of 802.15.4.
The related work showed that a filter at the center
frequency could thus increase packet delivery ratio
(PDR) for a periodic jammer at the center frequency
of the desired channel from under 5% to over 90%.
However, the previous approach fails if the at-
tacker adjusts the jamming center frequency beyond
the base-band filter. In Figure 3, we empirically show
how the single base-band filter yields worse perfor-
mance than an unfiltered receiver when the jamming
center frequency is shifted from the center of the
channel. In this work, we thus propose a technique
to mitigate the effects of the proposed jammer with
an arbitrary center frequency.
3 ADAPTIVE FILTER
SELECTION
As discussed in Section 2, it has previously been
shown that a filter can mitigate a jamming attack with
a center frequency at the center of the channel. In
this work, we allow that attacker to modulate to an
arbitrary center frequency. We consider two types of
attackers: one chooses its center frequency at the be-
ginning of the attack and never changes it, and the
other periodically changes its center frequency.
To mitigate such attacks, we propose to redesign
the 802.15.4 receiver as shown in Figure 4. In this
figure the light colored blocks represent the normal
802.15.4 receiver and the dark blocks represent pro-
posed modifications to the receiver, which include
filter-banks on the in-phase (I) and quadrature (Q)
channels as well as a controller to select the filter.
Our research hypothesis is that there exists a set
of filters Φ such that at least one filter φ ∈ Φ can al-
low a receiver to achieve a high PDR under a periodic
jamming attack at any center frequency assuming the
attack is of equal power to the legitimate signal. We
consider this through an empirical study in Section 6.
Since 802.15.4 uses a 2 byte CRC we propose
using information derived from these consistency
checks to keep track of packet delivery ratio for a
given time period. These calculations are performed
in the “Packet error detection” block of Figure 4.
Given a set of filters Φ and the PDR information,
we now have to derive the “Filter controller” block
from Figure 4. We propose a filter controller which
tests if PDR is higher then a threshold δ, if it is than
there is no need to attempt a defense, as there is no
effective attack being mounted. On the other hand, if
the PDR is lower than a threshold δ, there is either
an attack being mounted or environmental conditions
affecting the system. Since the system is already re-
ceiving a high amount of error, it is in the receiver’s
interest to try and mitigate this attack. To mitigate the
attack the receiver chooses a filter φ from the set Φ
and try receiving for a fixed amount of time τ. After τ
seconds the receiver then determines if PDR is again
under δ, if so it selects another φ ∈ Φ to try to increase
PDR. Once it finds a filter φ to maximize PDR, nor-
MITIGATION OF PERIODIC JAMMING IN A SPREAD SPECTRUM SYSTEM BY ADAPTIVE FILTER SELECTION
433