A Novel Detection Scheme to Mitigate the TD-LTE Interference

A Novel Detection Scheme to Mitigate the TD-LTE Interference

Jian FANG, Xiaopei ZHU, Biao HUANG, Xing guo ZHOU

Abstract--In this article, a novel detection scheme will be introduced into the CR system based on the Time Duplex-Long Term Evolution (TD-LTE) scheme, in which radar system is the primary user. If the CR nodes carry on the on-line energy detection to the radar signal, it will be subject to the interference originated from the TD-LTE

system's Base Station (BS) [11. We propose a new detection program employed only in the TD-LTE uplink timeslots. In this program, the interference, which is coming from the

TD-LTE system's User Equipment (UE) (2), may be weaker than it originated from BS. At the same time, the performance will be enhanced by accumulating multiple coherent detecting results. Through theoretical analysis and simulation, the feasibility of the new program has been verified.

1. INTRODUCTION

IN the past few decades, many new services need a growing number of spectrum resources with the rapid

development of wireless communication. But the limited frequency spectrum resources become the bottleneck of the development of wireless communication. How to improve the utilization of the spectrum has become a hot issue of the wireless communication research. Available spectrum resources are usually managed and controlled by the national unity. Various wireless communication systems are awarded the different frequency band to avoid disturbing. The massive researches of Federal Communication Commission (FCC) indicate that some frequency bands, such as the industry frequency, the science frequency, the medical frequency and some bands of land mobile communication around 20Hz [3], are too crowded. But some other frequency bands actually are frequently idle.

By making use of sub system opportunity-style primary system [4], Cognitive Radio (CR) technology can raise the frequency spectrum utilization effectively. Cognitive Radio system may detect various frequency bands continuously. If primary user is using this frequency band, secondary user will not be able to use the band; otherwise the secondary user may use this band pass-along message. In the course of secondary user, once discovered the primary user have begun to use this band, it will take an immediate end to the utilization of the band and switch to other suitable frequency bands.

In many frequency bands of the radar system [5], the spectrum utilization efficiency is low. Therefore, the introduction of cognitive technology has certain feasibility. Choosing a good program and algorithm to

This work is sponsored by National Science and Technology Major Project of China (2009ZX03007-004). Jian FANG, Biao HUANG and Xingguo ZHOU are with State Radio Monitoring Center, Beijing, China [email protected], [email protected], [email protected] Xiaopei ZHU is with Beijing University of Science and Technology, Beijing, China [email protected]

978-1-4244-8223-8/10/$26.00 ©20 1 0 IEEE 59

detect the spectrum is the key which the CR system may realize. Frequency spectrum sensation algorithm has been conducted the massive researches [6]-[8]. Four kinds of algorithm have introduced in a comprehensive article: matched filter detection, energy detection, cyclostationary detection and collaborative detection.

In this article, the TD-LTE system is the cognition user system while considering the radar system as the primary user system. In the detection process, the CR nodes have the possibility to receive the interference from the TD-L TE signals. As traditional energy detection algorithm is extremely sensitive to the noise and the interference, the detection performance will be decreased significantly especially in the serious condition. As we know, the interference generated by UE in uplink timeslots is smaller than it generated by BS in downlink timeslots. In addition, there is a certain distance between UE and the CR node. The interference power of the CR node which receives in the upward timeslots is far less than it in the downlink timeslots. Therefore, a novel detection scheme which based on uplink timeslots of the TD-L TE system is proposed.

To properly set the stage for the problem, we begin in section 2 with a brief review of the system model of this detection program. In section 3, the detection performance of this program is verified from the perspective of probability theory. Numerical results are presented in section 4. Finally, conclusions are made in section 5.

II. SYSTEM MODEL

The work scene of the CR system is shown in Figure 1 . Before the CR system working, the CR node carries on availability inspection to the channel. If surveying available channels, the CR node access to the monitoring mode. After coming on the monitoring mode, the CR node similarly needs to carry on the online examination to detect whether the primary user is transmitting signals. If it is, the TD-L TE system, as the secondary user, should withdraw from the channel and detect others again.

o TD·LTE

Base Station

o CRNode

Fig.l. System model

C Radar

In the initialization energy detection mode, influence is small the ordinary energy detection still can realize, because the interference is the thermal noise of the CR node. In the on-line energy detection mode, both the thermal noise of the CR node and the interference from the TD-L TE system are effect on the performances of the energy detection. In this article, the characteristic of TDD duplex mode is used. In the duplex mode, each transmission frame has an uplink timeslot and a downlink timeslot. It's shown in Fig.2 [9). In the TD-LTE system, the transmitting power of User Equipment (UE) is weaker than it of Base Station (BS). Executing on the detection in uplink timeslots in the on-line monitoring mode, it may greatly avoid serious downlink interference and obtain better detection performance. At the same time, in order to enhance detection performance, many uplink timeslots should be examined, and then carries on the coherent accumulation to the detection results.

Orel�o�amE �'�mT,'10n'l 4

I Oretll·fiam, liMT,'�� � �

lifiesli T.'153ro� !i)Ja)� �- �

Fig. 2. Frame Structure of one pair of switching points between downlink and uplink traffic timeslot

In considering the case of interference, the spectrum sensing based on energy detection can be seen as the binary hypothesis [ 10). It can be calculated by:

Ho. x(n) = men) + len) n = 1,2···N x(n) = s(n)+m(n)+l(n) n = 1,2···N

where x(n) is the received signal; men) is the thermal noise; len) is the TD-LTE uplink interference signal;

, (1)

sen) is the signal of radar. It obeys complex Gaussian

distribution with 0 mean and variance 8,; . a(n) = m(n) + len) (2)

a(n) obeys complex Gaussian distribution with 0 mean

and variance 8,; .

III. PERFORMANCE ANALYSIS

A. Detection failure probability and False alarm probability

Detection failure [II] refers that the CR node has not detected the radar signal while the radar signal appearing. Detection failure probability must be as low as possible not to interfere with radars. A value of Detection failure probability = I % is set as a target. The probability of detection varies with the power of the radar signal relative

60

to the detector threshold setting in the receiver - a higher power level increases the detection probability and it will bring about system efficiency losses.

False alarm refers that the CR node detects the interference pulse as radar signals while the radar signals is not present. False alarm probability is as far as possible low, so it will not reduce the efficiency of the TD-LTE. Consider the probability of false alarm, which has relationship with the size of the noise and interference, target of I %. When the signal to interference ratio (SIR) is low, it is likely to produce false alarm. Simultaneously the threshold establishment also similarly affects its size. If the detection threshold is enhanced as far as possible, the probability of false alarm will be cut down, but it will bring a great probability of detection failure.

In order to guarantee that the CR system correct and effective operation, selecting the appropriate threshold is particularly important.

B. Probability density function

The CR system's examination performance can be weighed by the received average power's probability density function [1 2). Through the probability density function and threshold settings can calculate the detection probability and the false alarm probability.

The noise and interference aJn),a,(n)···aN(n) are

complex Gaussian distribution. a,(n) , a,(n) , GN(n),isX2with' (--) + (--) + ... + (--) .

0" 0" 0"

T mea Deg _n = 2*----

T (3)

degrees of freedom, where T _mea is the measurement time ,and T is the sampling time.

Inspection function is given by:

(a'(n) + a:(n) + ... + a'(n») v = lOlg "

.,

n (4)

So cumulative probability density function (CDF) is given by:

F(v) = P(V <v)

n

U =P(lOlg-- <v)

n / 0'; = P(U <-1010) 8,:

I' n -=F, (-10") 7. (n) o,�

Probability density function is given by: v

(6)

�IOlOlnl0 v f(v)= F'(v)= c5; 10 'fx'(b)C

c5n2101O) (7) n

When the radar signals are caught, the received

signals PI (n), P, (n) . . · PH (n) are complex Gaussian

distribution.

(p,(n)), +(p,(n)), + ... +(PN(n))'is X2 with: is" is" is"

T radar Deg _n = 2* ----

T (8)

degrees of freedom, where T _ radar is the radar pulse width.

Probability density function of T radar is

n ---- 1010lnlO 8: + 8;

j(p) = F (p) = ----'----j 2 ( n

1010) I (n) t5/� + a; 10

C. Performance Analysis

(9)

Base on the noise and the signal probability density function in the single examination situation, the detection performance as fellow:

Probability of detection failure can be calculated by:

Ph _df = P(Radar _l ev e l < R_th)

= L:_th f(p )dp ( 10)

Probability of false alarm can be calculated by:

Ph _fa = P(Noise_level > R_th) = S::J(v)dp ( 1 1 )

where R th is detection threshold. Noise _level and Radar _level are the values of the

corresponding level. In order to improve the capability, cumulative

detection performance of several detects as fellow: Probability of detection failure is given by:

� _df _m = (P(Radar _level < R_th))'"

=

(1:-th f(p)dp r

Probability of false alarm is given by:

� _fa_m = 1- (1- P(Noise _level> R_th)r ( R th )m = 1- L- f(v)dp

where m is the number of detection.

IV. SIMULATION AND DISCUSSION

( 12)

, ( 13)

In order to appraise the performance of this plan, the curve of detection failure probability and false alarm probability are compared under different threshold [ 1 3).

In TABLE I, these values are used in the calculations. Signal measurement time is equal to the width of pulse

radar. Sampling interval is less than 1 0MHz in the TD-LET system. iSp _dB is the received power of the CR node. When the TD-LTE and radar system coexist, is. _ dB is the marginal value of interference power. -70dBm and -80dBm respectively are the power of the CR node, when the TD-LTE is in the downlink and uplink mode.

In Fig.3, it can be seen that the settings of threshold may influence the detection probability and the probability of false alarm. If the distance between the two

61

curves farther, the CR system's signal to noise ratio (SNR) is greater and the controllability of threshold is better.

TABLE I VALUES USED IN THE CALCULATIONS

Symbol

T mea

T radar

T m

t5n t5p

0.7

Value

41ls 41lS

O.IIlS 111 0

-85dBmI-70dBm

-77dBm

.*". Noise �- Radar and noise 0.6 ____ .l-____ --!-____ -l ___ .L-_____ -----'

0.5

0.4 "-0 a.

0.3

0.2

0.1

-100

- - - - i -i -l-� - - J -----i-----I _ot�_ I Z . I I I _;:: I + t I I

- - - - � - ;-� - � 1 1 - - � - --

- - : - - - --

I +-:- I I I I I I *;c I T I I I

- - - - T - :! - -,- -fj 1 - -,-- - - -1- - ---I -,- -t- I . I I I -+ ... I I I I

- - - - � -H -� ; -1--�- ----:-----, ,- t '1" ,

- - - - � --� i:- � .. i- - �- ----:-----I I I I

-90 -70 Detected power

-60 -50 -40

Fig. 3. The power spectral density of detection power

100 �7���''''��!''''II!!''l!''!!!I!!!!''!!!III!<,'�_� __ !!,!,,!� Detection failure

Fig. 4. Detection performance in the TD-L TE uplink timeslot

-75 -70 -65 -60 -55 -50 Power of threshold

Fig. 5. Detection performance in the TD-L TE downlink times lots

If the values of the detection failure probability and the false alarm probability are set to 1 % as targets, it can be seen that in the uplink timeslots the existence of an appropriate threshold and its adjustable range about 4 dB in Fig.4. The downlink timeslots examination is unable to satisfy the targets of both the detection failure probability and the false alarm probability simultaneously in Fig.5. The curve is quite sharp therefore it is quite sensitive to the SNR. The curve rapidly intersects with the increase of interference.

I· Detection faiture -.- False alarm

Fig. 6. Detection performance on 5 occasions in the TD-LTE downlink timeslots

Fig. 7. Detection performance on 10 occasions in the TD-LTE downlink timeslots

In Fig.6 and Fig.7, we can see that the distance between detection failure probability and false alarm probability is farther. When the detection failure probability and false alarm probability are 1 %, there are more than 5dB adjustable threshold. Many occasions detection performance is much better than the performance of single detection. Compared with Fig.6 and Fig.7, the capability has not obtained a large promotion. In considering of the computation load, it should choose appropriate numbers of detection occasion.

62

V. SUMMARY OF SIMULATION RESULTS

Energy detection's performance is highly susceptible to uncertainty in estimation of the fluctuating noise power and interference power, especially online detection in the scenario stated in the paper. It is shown that the scheme, utilizing detection in uncontinuous uplink timeslots of the TD-LTE, obviously mitigates the interference and enhances the performance of accumulating multiple detections. We demonstrate the improvements on spectrum sensing performance in various parameters through theoretical and simulation analysis.

REFERENCE [1] Base Station (BS) radio transmission and

reception,(Release8)3GPP.TS36.104.V8.2.0(2008-05) [2] User Equipment (UE) radio transmission and

reception,(Release8)3GPP.TS36.101.V8.2.0(2008-05) [3] S. Haykin, "Cognitive radio: brain-empowered wireless

communications," IEEE Journal on Selected Areas in Communications, voL 23, no. 2, pp. 201-220, Feb. 2005

[4] Federal Communications Commission, "Spectrum Policy Task Force", Rep. ET Docket no.02-135, Nov. 2002

[5] ITU-R M.1464, Characteristics of and protection criteria for radionavigation and meteorological radars operating in the frequency band 2700-2900MHz

[6] D. Cabric, S. M. Mishra, R. W. Brodersen, "Implementation issues in spectrum sensing for cognitive radios," Conference Record of the Thirty-Eighth Asilomar Conference on Signals, Systems and Computers, 2004. 772- 776

[7] 1. Yucek, H. Arslan, "A survey of spectrum sensing algorithms for cognItive radio applications," IEEE Communications Surveys & Tutorials, voL 11, no. 1, First Quarter 2009, pp. ll6--l30

[8] S. Haykin, D. 1. Thomson, 1. H. Reed, "Spectrum sensing for cognitive radio," Proceedings of the IEEE, vol 97, no. 5, May 2009, pp. 849-877

[9] Physical layer aspects for evolved Universal Terrestrial Radio Access (UTRA),(Release 7)3GPP TR 25.814 V7.LO (2006-09)

[10] T ANORA R, SAHA I A. Fundamental limits on detection in low SNR under noise uncertainty [C] II Proc of the Wireless Comm' 05 Symposium on Signal Processing. 2005: 4642469.

[11] S. Appadwedula, V.V. Veeravalli and D.L. Jones, "Energy-efficient detection in sensor networks", IEEE Journal. Selected. Comm., voL 23, no. 4, Apr. 2005

[12] Sheldon Ross , " First Course in Probability," [13] W. Y. Yang, W. Cao, 1. S. Chung, and J. Morris, APPLIED

NUMERICAL METHODS USING MATLAB, John Wiley &Sons, 2005.