Z. Nikolova, V. Poulkov, G. Iliev, G. Stoyanov

1. NARROWBAND INTERFERENCE CANCELLATION IN MULTIBAND OFDM SYSTEMS Z. Nikolova, V. Poulkov, G. Iliev, G. Stoyanov Dept. of Telecommunications Technical University of Sofia BULGARIA e-mail: zvv@tu-sofia.bg

2. Multiband OFDM (UWB): • Technology for short range high data rate communications, combining OFDM and frequency hopping. • Occupies a very wide frequency band and low transmission power; • UWB systems are subject to different types of narrowband interferences, which could deteriorate strongly and even block communications; • NBI suppression is of primary importance for these systems.

3. Multiband OFDM

4. NBI mitigation • NBI avoidance methods. Based on avoiding the transmission over frequencies with strong narrowband interferers; • Cancellation – suppression methods. Aim at eliminating the effect of NBI on the received UWB signal.

5. Scheme for the suppression of NBI • How? • Via adaptive complex filtering using the LMS algorithm to adapt to the central frequency of the NBI. • Why is possible? • Compared with the desired wideband signal the interference occupies a much narrower frequency band, but with a higher power spectral density; • UWB signal has autocorrelation properties quite similar to that of AWGN, so filtering in the frequency domain could be applied.

6. Scheme for the suppression of NBI

7. c f b + Variable complex digital filter circuit derivation The composite multiplier, containing b is derived: First-order complex coefficient transfer function Second-order real coefficient transfer functions All of them of BP type, describing a complex digital filter section, which is variable with respect to both the BW (by changingofc) and the central frequency (by changingofq).

8. Variable complex digital filter circuit derivation Complex second-order digital filter section Real first-order digital filter section

9. Variable complex second-order filter Magnitude and group-delay responsesof variable BP complex second-order filter for different values of the central frequency

10. Variable complex second-order filter Magnitude, phase and group-delay responses of variable BP complex second-order filter for different values of the bandwidth

11. ADAPTIVE ALGORITHM eR(n) xR(n) SECOND-ORDER COMPLEX FILTER yR(n) yI(n) xI(n) eI(n) + + Adaptive complex second-order filter The cost-function is the power of BS filter output signal: where The Least Mean Squares (LMS) algorithm is applied to up­date the filter coefficient responsible for the central frequency as follows: mis the step size controlling the speed of the convergence; (*) denotes complex-conjugate; y(n) is the derivative of y(n) with respect to the coefficient subject of adaptation. Block-diagram of a versatile adaptive complex narrowband filter For the BP filter we have the following real and imaginary outputs: and - when the input signal is - when the input signal is For the BS filter we have - real and imaginary outputs: and

12. Performance evaluation of the NBI suppression scheme via simulations relative to baseband: • Assuming standard MF-OFDM receiver with 3 subbands (as proposed MB-OFDM in the 3.1 – 4.8 GHz band) and IEEE 802.15 3a Channel Model 1. • The NBI interference is modulated with a random frequency appearing in one subband; • Soft decision 4-bit Viterbi decoding was used, thus recovering some of the lost data due to the frequency domain excision.

13. Results: • The simulations showed that for one and the same BER such NBI cancellation scheme gives more than 2 dB improvement for signal-to-interference ratios below 8 dB. • Drawback of the scheme: • Frequency excision is performed over all OFDM symbols (dehopped over all subbands), nevertheless the fact that the NBI appears only in one.

14. Better if: • Frequency excision is performed not over all symbols but only on those affected by NBI. • Proposed solution: • Instead of one complex adaptive filter section a filter bank with corresponding switching capabilities and a number of ACF equal to the number of subbands is implemented. • Corresponding ACF to be switched accordingly with the hopping of the carrier frequencies and will process only the OFDM symbols appearing in one and the same frequency sub-band.

15. Analysis: • Behavior of an ACF, composed of three second-order complex filter sections. • Input interfering signal is composed of three complex sine-signals with frequencies f1=0.25, f2=0.2 and f3=0.15; • Learning curves show that our filter bank is able to detect and to track the input complex sinusoids and can be successfully used for their cancellation.

16. Adaptive complex second-order filter Learning curves for second-order adaptive complex filter applying different step size

17. Adaptive complex second-order filter Learning curves of an ACFB consisting of three second-order complex filter sections

18. Advantages of the ACF: • Low computational complexity; • Fast convergence (less than 100 iterations in the example); • Convenience for implementation with CORDIC processors; • The very low sensitivity of the initial LP section ensures a high tuning accuracy.

19. Another important advantage: The proposed NBI cancellation scheme is that the adaptive complex notch filter section has also a bandpass output used for monitoring the NBI and switching of the ACF in cases when the NBI vanishes or is reduced to a predetermined level.

20. THANK YOU!