An Efficient Security Method Based on Exploiting Channel State Information (CSI)


Hamamreh J. M., Furqan H. M., Ali Z., Sidhu G. A. S.

15th International Conference on Frontiers of Information Technology, FIT 2017, Islamabad, Pakistan, 18 - 20 December 2017, vol.2017-January, pp.288-293 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 2017-January
  • Doi Number: 10.1109/fit.2017.00058
  • City: Islamabad
  • Country: Pakistan
  • Page Numbers: pp.288-293
  • Keywords: Channel State Information, Security
  • Istanbul Medipol University Affiliated: Yes

Abstract

A channel amplitude quantization method that can effectively quantize the channel response using just one single threshold value is proposed in order to extract a random manipulating sequence with good secrecy properties. Specifically, a Time Division Duplex (TDD) wireless system is considered over independent identical distributed (i.i.d.) Rayleigh fast fading channel, where potential passive eavesdroppers (Eves) can only estimate their own channel and have no knowledge about CSI between legitimate communication parties. The transmitter (Alice) is only aware of the CSI of the legitimate user (Bob). Particularly, the proposed security technique takes the bits of the transmitted data packets and manipulate them with a logical vector that characterizes the channel randomness based on the estimated CSI gain. The process of manipulation is implemented on a bit level basis using an XOR operation exactly before modulation process. The same XOR operation is implemented after demodulation process on the detected bits to extract the concealed bits. The obtained simulation results show that the employment of such mechanism can ensure data confidentiality. Furthermore, the simulation results are extended to include the effect of the selected quantization threshold on the BER performance of Eve as well as the amount of information leakage to its side. It is shown that security gap region between Bob and Eve is made very large over all expected Signal to Noise ratio (SNR) values despite the small degradation in the bit error rate (BER) performance of Bob because of the expected channel estimation errors due to noise.