Network-controlled physical-layer security
The challenge
A society without wireless devices is unthinkable. Were reliant on wireless communication technologies to exchange personal and, sometimes, confidential data. The broadcasting nature of the wireless medium makes exposure to eavesdroppers a potential threat.
So far, this threat has mostly been mitigated by encrypting the wireless link and the information transmitted. Such a solution assumes that eavesdroppers lack the computational resources and knowledge of the network parameters to break the encryption. While this assumption still holds for many scenarios, eavesdroppers' capabilities are rapidly improving, particularly with the availability of quantum computers on the horizon.
has been widely recognised as a complementary and, sometimes, alternative approach to encryption. PLS limits the amount of information that can be intercepted by assuring the signal contains so much noise at the electromagnetic level, that its impossible for eavesdroppers to decode any data, thus achieving perfect secrecy.
Several techniques have been proposed to implement PLS, falling into the categories of channel coding, channel control and power control. So far, these techniques remain limited to the information theory domain, without practical implementations.
The solution
We have demonstrated that PLS can be realised using off-the-shelf equipment by tackling the problem at the network level. The idea is that a wireless network typically contains not just one wireless access point (AP), but many APs to which a legitimate station could possibly connect.
Using a relatively new enhancement of Software-Defined Networking (SDN) specifically for wireless networks, called spectrum programming, its now possible to execute intelligent AP selection algorithms in a completely transparent way to the connecting station. We investigated two such algorithms in our earlier work. We had the legitimate station always connect to the least beneficial AP to the eavesdropper, or the AP was selected that maximised the secrecy capacity for the legitimate station.
The secrecy capacity is the maximum capacity a legitimate station can achieve under the condition of full secrecy while connected to a given AP. We also introduced a novel secrecy capacity optimisation algorithm in our recent in which we combine intelligent AP selection based on maximising secrecy capacity with the addition of (FJ) by the not-selected AP. We showed that providing such a FJ signal to the eavesdroppers significantly improves secrecy in the network beyond what can be achieved with intelligent AP selection.
The impact
Our results are still recent and we have only just started the dissemination process. We have presented our solution at various conferences and posted a demonstration video on Were extending the work with improved modelling and larger implementations, as such proving the scalability and portability of our solution, and were inviting industry partners to join us in that journey.
The impact this work may have on society is significant. Hundreds of researchers worldwide have been working on PLS over the past two decades, having contributed thousands of papers, because the potential of PLS is so strong: it extends the security of wireless networks well beyond the domain of cryptography. We have proven that PLS can be implemented in a cost-effective way, which may very well be the final push this technology needs for adoption by the market.
For us to be able to demonstrate that current networks are ready for cost-effective deployment of Physical-Layer Security is a true victory of joint R&D in information theory and wireless network engineering.
Research Lead, Professor Parastoo Sadeghi, UNSW School of Engineering & Information Technology
