Signal-to-Noise Ratio of CMOS-Based FMCW Sensor System…

The SNR of the proposed FMCW sensor system was characterized theoretically before performing additional measurements. Additionally, the calculated data were confirmed by the experimental results of a field test. Figure 6.6 presents the setup for vehicle detection using the proposed sensor system.

The sensor, which is a road-side unit, was installed above the ground at the height h. The vehicle occupancy in multiple lanes was detected by performing a range measurement with the FMCW sensor system. The total width of the multiple-lanes is denoted by D in Fig. 6.6, and represents the maximum coverage of the FMCW sensor. The values of h and D were set to 3.0m and 30m, respectively, in the field test. The building blocks of the FMCW sensor system, which had a CMOS transceiver, a digital signal processing unit and two antenna arrays, were integrated within a metallic enclosure before experiments. The width, length and height of the enclosure were 20cm, 22cm and 12cm, respectively, according to Fig. 4.31. Two antenna arrays were installed on top of the enclosure, and covered by a radome, which is a Teflon sheet with a thickness of 2.0mm. The H-plane of the antenna array was orthogonal to the traffic lane, and slanted at an angle of 35° by mechanical rotation, as indicated in Fig. 6.6.

The configuration helped the FMCW sensor to suppress the Doppler effect of the moving vehicle.

The csc²θ type radar was constructed from the present antenna array with the H-plane pattern, and the appropriate sensor height and mechanical rotated.

The major capability of a csc²θ type antenna pattern is performing the multiple-lanes measurement. The antenna pattern thus has a similar function to

the sensitivity time control (STC) of the IF filter, of providing a spatially uniform SNR within the range measurement. The proposed antenna arrays not only realized the nearly spatially uniform SNR to compensate the power level at farther section, but also achieved high isolation, thus suppressing heavy interference in the near field.

Fig. 6.6 The FMCW sensor adopted the multiple-lane vehicle detection of TMS. The H-plane antenna radiation is orthogonal to the moving vehicle

The power of the electromagnetic waves that propagate between the sensor and the single vehicle was estimated from the range measurement presented in Fig. 6.6 by the following well-documented equation: [49]

4

where S denotes the receiving power at the output of the receiving antenna;

P_TX represents the transmitting power at the input of the transmitting antenna; G_T

Elevation angle θ= 0°

θ is measured from broadside

cross section (RCS) of the vehicle at the operating frequency, and R represents the effective distance between the sensor and the vehicle [49].

According to stated electrical characteristics of the proposed CMOS transceiver and antenna arrays in chapter 3 and 4, the operating frequency was 10.5 GHz, and the corresponding wavelength in the air was 2.86cm. Since the H-plane main-beam of an antenna array was aimed at the vehicle by making mechanical adjustments, the antenna gains for both transmitting and receiving paths were calculated from the curves plotted in Fig. 4.25. Additionally, an irregular metallic body was applied as the reflected target with an effective RCS (σ) of about 0dBsm at 10.5 GHz by the user manual and the RCS of the irregular metallic body be equivalent to a spherical object with a radius of 1.0m [49]. The output power of the CMOS FMCW transceiver in Fig. 3.6 was

−18.2dBm at 10.5GHz, rising to 0dBm at the input of the transmitting antenna through an external PA. Therefore, the reflecting power of single target at the input of the CMOS transceiver was about −87dBm at D=30m. The conversion gain from LNA input to the mixer output at the IF output port in the receiver path was 0.2dB. The 49dB IF amplifier was found to be adequate for enabling the DSP unit to carry on signal processing in the presented sensor system.

Significantly, the leaky-mode antenna array eliminates the R⁴ equalizer, thus compensating for the R⁴-dependence of returned signal power. The property is similar to the csc² type antenna of the surveillance radar by military application.

The most important contribution of this study is to present a sensor system that supports the almost equal to echo power for range measurement of multiple lanes, using the H-plane antenna pattern in Fig. 4.25. The gain factors GT and GR

in (6-1) can be extended to a cross-section of the H-plane pattern in Fig. 4.25 as

G_T(θ) and G_R(θ).The gain of G_IF is sum of a mixer and an amplifier was added to radar equation in order to compare the measured data directly. A new radar equation was derived from (6-1) with reference to Fig. 6.6, which shows the antenna arrays.

where θ denotes the elevation angle that measured from the broadside; β represents the elevation angle set by mechanical rotation, and GT(θ+β) and G_T(θ+β)indicate the transmitting and receiving antenna patterns, and are functions of θ and β. Restated, the antenna pattern functions are shifted through an angle of β. The parameter h is the height of the sensor box containing the antenna arrays, the radio frequency module and other circuits. The function h⋅cscθ was substituted for R in (6-1) using the trigonometric relationship in Fig.

6.6. The term R is a function of h and β, and is given by

Equation (6-3) and new parameters in (6-1) yield the modified radar equation, (6-2). The receiving power S is a function of θ, β and h. Equation (6-2)

Since the angle of maximum gain of the H-Plane in Fig. 6.6 is 56°, and the H-plane pattern has a special envelope region at about 60°−70°, the echo power decay 1/R⁴ can be compensated. The study presents a simple approach for maintaining a near-constant echo power in the range measurements of the FMCW system, by tuning the angle (β) by mechanical rotation to the envelope region of the antenna pattern, thus projecting it exactly onto the multiple-lanes.

Equation (6-2) was employed to estimate the optimal angle of mechanical rotation, yielding a uniform reflected power response for range measurement. In Fig. 6.7(a), the power distribution of the beat frequency was varied, and the mechanically rotated angle was varied from −5° to −55° in steps of 10°. The antenna arrays were 3m above ground. The most uniform power responses were between −45° and −55°. The region of highly uniform power distribution was expanded as in Fig. 6.7(b), where the angular interval was 2°. Accordingly, the envelope curves of the H-plane pattern between −45° and −55° were adopted to measure the distance, compensating for the loss, which is given by the inverse of the fourth power of the distance. However, the near-uniform power response was associated with greater power decay at shorter ranges, as indicated in Figs.

6.7 (a) and (b). The near-field interference of most transceivers resulted from leakage from the LO port, or coupling of the transmitter and the receiver, which always occurred at the short range.

The shorter range power decay can be cancelled by the near field interference. If a system requires uniform spatial power at the short range, then the height of the sensor can be reduced to meet the requirement. Figure 6.7 (c) shows the variation of the power distribution of the beat frequency with the height of the sensor from 1m to 5m in steps of 1m, at an angle of mechanical

rotation of 50°. Figure 6.7 (c) reveals that a higher sensor was associated with more power decay at the short range, but a more uniform power response at a longer range. A lower sensor yielded a better power distribution at the short range, but a non-uniform power response at the long range.

5 10 15 20 25 30

Power of beat frequency (dBm)

Elevation Angle of Ant.

-5degree

Power of beat frequency (dBm)

Elevation Angle of Ant.

-45degree

5 10 15 20 25 30

Power of beat frequency (dBm)

1m 2m 3m 4m 5m

Fig.6.7 (a) Estimation of echo power distribution with fixed sensor height fixed at 3m, and angles of rotation from −5° to −55°. (b) Magnified figure (a) from −45° to −55°. (c) Fixed angle −50° and variation of height of sensor from 1m to 5m.

The noise of the completed sensor system of the proposed X-band CMOS chip was characterized by performing time-domain system simulations by a commercial software package, ADS2006A. The system simulations represent the building blocks shown in Fig. 1.1 as behavior models. The parameters of each model were carefully set up based on the measured results of the corresponding device.

Fast-Fourier transformation was performed on system simulation results.

The noise value of the CMOS transceiver after FFT was 9.2 dB, which is 0.8dB less than that obtained from the measurements in Table 3.1. Therefore, the noise power of the completed radar system is given by

F d

N

kTB N

P =

(6-5) where k denotes the Bolzmann constant, and T represents the ambient temperature (K) [49]. B_d denotes the bandwidth of the range filter. The temperature was set to room temperature, and Bd=3.9 KHz, which approaches

the bandwidth of frequency resolution. The noise figure N_F in the presented CMOS transceiver was 10dB. Therefore, the total noise power of the sensor system was −128dBm.

For convenience of range measurement and uniform spatial power distribution, the angle producing the largest gain of the H-plane pattern in the farthest lane was used, enabling the optimal uniform power response to be obtained easily. The trigonometric relationship of Fig. 6.6 was also employed to derive the angle ψ of maximum gain at a particular distance, which is given by

) ( tan ¹

h

− D ψ =

The rotation angle is given by

β

=

ψ

−56° then

°

−

= tan

⁻¹

( ) 56 h

β D

(6-6)

Table 6.1 lists estimated optimal angles of mechanical rotation for between 10m and 60m, as determined from (6-6). If the sensor height is fixed at 3m, then Table 6.1 yields the distance of the farthest lane that corresponds to the angle of mechanical rotation. Table 6.1 covers distances of the farthest lane from 20 m to 60 m, corresponding to angles of mechanical rotation of between −47° and −54°.

These angles are also associated with the best uniform spatial power distribution.

Table 6.1 Farthest lanes corresponding to given angles of rotation

D 10m 15m 20m 30m 40m 50m 60m

β -39° -44° -47° -50° -52° -53° -54°

The general definition of SNR was measured on the receiving port and calibrated the gain of LNA, mixer and IF AMP. The SNR equation (6-7) was formed by combining (6-2) and (6-5). The SNR parameters are defined in the preceding paragraph

The theoretical SNR was converted to dB form by subtracting the power of the receiving signal from the total noise power.

SNR (dB) = 0 dBm + dB(GT(θ+β)) + dB(GR(θ+β)) – 30 dB + 0 dB – 33 dB – 40 log (3⋅cscθ) + 128 dB (6-8) The SNR of the proposed sensor system was experimentally characterized by performing extensive field tests. The results were then compared with the theoretical data using (6-8). Figure 6.8 illustrates composite plots of SNR versus distance to the comparisons angles (β) of mechanical rotation of −35° and −50°.

The two groups of curves indicate excellent agreement for distances of 10–30m.

Near-field interference was prevalent at less than 10m, but was canceled by the envelope of the proposed H-plane pattern and the isolation of the presented CMOS chip. At angles of −35° or −50°, the SNR of the CMOS-based FMCW sensor system increased from 25dB to 42dB as the distance increased from 5m to 30m. The unique characteristic results mainly from the radiation pattern of the

leaky-mode antenna arrays. The H-plane radiation pattern from 5.7° to 31°

covered the multiple-lanes distances of 5–30m, based on (6-4). The antenna gain of the FMCW sensor system was inversely proportional to the square of the distance. The unique radiation pattern compensated for the signal degradation, as predicted by (6-3). At an angle of −50° , which was the angle of maximum power when the H-plane was aimed at the farthest lane, the power of the reflected signals was maintained the most constant for range measurement, yielding a nearly constant value of SNR when using the proposed FMCW sensor design. The SNR measurement result shows that the structure of the presented FMCW sensor has a good STC function for measuring the range of TMS.

5 10 15 20 25 30

D (m) -60

-40 -20 0 20 40 60 80 100

SNR (dB)

Measured(-35°) Calculation(-35°) Measured(-50°) Calculation(-50°)

Fig. 6.8 Signal-to-noise ratio (SNR) of CMOS-based FMCW sensor in Fig. 6.6.

在文檔中 調頻連續波感測器射頻前端設計與系統整合驗證 (頁 101-111)