IEEE - Aerospace and Electronic Systems - February 2020 - 14

Interference and Intrusion in Wireless Sensor Networks

Figure 6.
DPX visualization of the ISM RF Spectrum during the benchtop experiments. (a) DPX image showing the ZigBee signal at 2475 MHz.
(b) DPX image showing coexistence of ZigBee with 802.11 WiFi. (c) DPX image showing coexistence of ZigBee with 802.11 WiFi and
Bluetooth. (d) DPX image showing the spectral environment when ZigBee packets were lost.

Furthermore, other attacks (see the "Network-Layer
Attacks" section) on the upper layers of the protocol stack
can be detected by analyzing both the routing process and
network properties. However, detection can become skewed
if the attackers are subtle about their operation. For example,
monitoring network operation and implementing a sinkhole
attack that sporadically drops critical packets, which may
mimic PER levels resulting from a noisy environment, or a
wormhole attack tunneling only one of every N packets.
With the threat of legitimate nodes being captured, detection
and security mechanisms need to account for malicious
nodes in the network. Finally, it is clear that an attacker is
either an "outlaw" who breaks spectrum laws (excessive
radiated power) or a "noncompliant" operator who adheres
to broadcast power limitations but simply refuses to follow
protocol operation (service refusing to give up resources).
Additionally, each layer, from the PHY upwards is vulnerable and it is clear that intrusion detection and security are
complex processes and cannot simply focus on one aspect
but, rather, must examine the whole process from transmission to reception, and everything in between.

FUTURE DIRECTIONS
Security and intrusion detection, both intentional and unintentional, is integral for the future of successful WSN
14

deployments, especially in critical applications, like aerospace, space-based WSNs, IoT, and using nanosatellites as
relay nodes. Security and IDSs cannot simply focus on particular attack strategies, they must also consider coexistence
issues as intrusions and need to recognize that hostile noisy
environments exist. Due to the flexible topology, open interface and power limitations of these WSNs, this is a complex
challenge and needs to be solved to allow WSNs to be used
in safety critical applications and to safely exploit COTS
devices and standardized protocols. Based on the aforementioned pillars of security (vulnerabilities, requirements,
attacks, and defenses) and attack discussion, future work lies
in security development in terms of intrusion detection, both
intentional and unintentional, and mitigation. Attack effects
on WSNs and the associated signals and analysis of why
each specific security technique is used will be beneficial.
Additionally, the data from the PHY layer has potential to be
investigated as the radio architectures are usually very similar between wireless standards and so, by concentrating on
the PHY symbol stream and related measurements, the possibility of designing a transferable solution exists. This future
work entails both a reactive and detection approach, which
has potential to be environment specific. As the wireless
channel is a nonlinear phenomenon, an approach which can
model nonlinearities and adapt to new models is recommended. From this perspective, a feature/featureless-based
machine learning algorithm focused on received samples in

IEEE A&E SYSTEMS MAGAZINE

FEBRUARY 2020

IEEE - Aerospace and Electronic Systems - February 2020

Table of Contents for the Digital Edition of IEEE - Aerospace and Electronic Systems - February 2020