Signal Processing - November 2016 - 115

at the transmitter side. The proposed algorithm allocates the
channels to the users in such a way as to maximize a utility function that captures the joint uplink-downlink QoS
requirements of the users.

Physical layer security in wireless networks

establish secure communications for multiple source-destination pairs from a malicious eavesdropper, through the assistance of multiple friendly jammers. To enhance the overall
network secrecy rate [43], source nodes are willing to provide
a certain amount of monetary compensation to jammers in
exchange for jamming power. The matching algorithm in [9]
determines both 1) the matched source node-friendly jammer
pairs and 2) the exact amount of money transfer that motivates
both source nodes and friendly jammers to cooperate, such
that the final matching is stable and maximizes the sum of
utilities of all source nodes and friendly jammers, referred to
as the network social welfare. The framework is a dynamic
matching with transfer and the proposed matching algorithm
converges to a competitive equilibrium.
Figure 3 shows the physical layer security wireless network
model comprising multiple source node S-destination node D
pairs, with each pair using one spectrum band with the same
size. There is a set of multiple friendly jammers J and one
malicious eavesdropper node in the network. For simplicity,
the friendly jammers are all equipped with multiple antennas,
while the eavesdropper, source nodes, and destination nodes
are equipped with a single antenna each.
Friendly jammers compete with each other to assist a particular source-destination pair by creating sufficient interference to the eavesdropper. In exchange, the source-destination
pair will pay a monetary amount to the jammer. For simplicity,
it is assumed that during each time period, each source node
will be assisted by only one friendly jammer.
First, access points (APs) that are not matched make a
price offer to the unmatched source nodes. Each source node
then chooses the jammer that provides it with the highest positive utility. Then, each jammer will decide if it wants to match
with the source nodes. There are three possibilities. The first

The security of a wireless communication link has always been
of great importance. The broadcast nature of the wireless transmission medium makes eavesdropping easy, and anyone within
communication range can receive and possibly decode private
transmission signals. Moreover, higher-layer security key
distribution and management may be difficult to implement and may be vulnerable to attacks in some environments, such as ad hoc or relay networks, in which transceivers
may join or leave randomly [40]. Alternatively, physical layer-based security explores the characteristics of the wireless
channel to improve wireless transmission security.
The use of a friendly jammer to facilitate the degradation
of the source to an eavesdropper channel has been considered (see, e.g., [41]). This is achieved by a friendly jammer
transmitting a jamming power signal, which has the effect
of decreasing the signal-to-noise ratio at the eavesdropper.
This approach is often referred to as cooperative jamming.
With the additional degrees of freedom provided by multiantenna systems, friendly jammers can generate artificial
noise to degrade the channel condition of the eavesdropper
while maintaining little interference to the source nodes.
Most existing literature assumes perfect channel knowledge from the source to the eavesdropper. This is valid if
the eavesdropper is part of the communication system. For
example, in [42], a scenario was considered in which the
receivers eavesdrop on the message intended for other receivers. However, in some cases, the eavesdropper may not
be a user of the system, and obtaining
the eavesdropper channel information
is difficult. The algorithm that faciliSources
Destinations
tates using the eavesdropper needs to
S1
have acceptable performance without
D2
.
D1
perfect channel knowledge.
.
S2
.
Another requirement is to select the
DN
best friendly jammer for each source-
.
.
destination pair when there are multiple
Safe Communications
(S1-D2)-Friendly
.
SN
Jammer Matching
friendly jammers and multiple source-
destination pairs. The final selection
Eavesdropping
needs to be stable to guarantee a longterm secure communication. There are
J5
J1
Jamming
some centralized [43], [44] and distribSignal
J2
uted [45], [46] methods based on game
theory that have been proposed to handle
the interaction between the source nodes
Malicious
J4
J3
Eavesdropper
and the friendly jammer. However, they
did not cover a general scenario with
Normally as a
Friendly Jammers
multiple source-destination pairs and
Member of the Network
multiple friendly jammers.
In [9], a matching theory-based FIGURE 3. The physical layer security wireless network model with multiple friendly jammers equipped
distributed algorithm is proposed to with multiple antennas.
IEEE Signal Processing Magazine

November 2016

115

Table of Contents for the Digital Edition of Signal Processing - November 2016