Analysis of Concatenated Waveforms and Required STC Jian Wang*, Eli Brookner**, and Mark Gerecke* * Raytheon Canada Limited 400 Phillip Street, N2L 6R7, Waterloo, Canada phone: + (1) 519-885-0110 ext. 768, fax: + (1) 519-885-8672, email: Jian_Wang@raytheon.com ** Raytheon Company 528 Boston Post Rd., MA 01776, Sudbury, USA phone: + (1) 978-440-4007, email: Eli_Brookner@raytheon.com Abstract—In modern surveillance radar, pulse compression is applied to achieve long range coverage while maintaining both low transmit peak power and desired range resolution. When there is a requirement to increase the radar’s range without a corresponding increase in the transmitted peak power it is required that a longer uncompressed pulse be used. As a consequence the blind range, associated with blanking the receiver whilst transmitting the pulse, will increase and short range coverage will be lost. In this paper we initially present a system of concatenated waveforms that provide simultaneous radar coverage for both near and far ranges. The solution incorporates a combination of longer sub-pulses for far range and shorter sub-pulses for near range. Although the concatenated waveforms have a number of attractive attributes, they present a challenging problem in that different sub-pulses will reflect back from different ranges and be received at the same time. Therefore since each sub-pulse is not separable at the RF front end a more complex sensitivity time control (STC) is required. In the second part of this paper we analyze this phenomenon and propose a STC scheme to address the issue. Index Terms— STC, Concatenated Waveform, Receiver Saturation. I. INTRODUCTION In modern surveillance radar systems, both high range resolution and long range coverage are important features [1]. High range resolution may be obtained by a short duration pulse. Since the sensitivity of a radar relies on the energy contained in the transmitted pulse, far range may be reached by employing a long duration pulse and/or high transmitter power. Simply increasing the radar pulse length has the side effect of degrading the range resolution. Alternatively short pulses require higher transmitter power, which translates to higher cost and lower reliability [2]. Pulse compression techniques [1-3] allow reduced transmitter power whilst still maintaining long range and high resolution. The reason is that the range resolution does not necessarily depend on the duration of the pulse but on the bandwidth of the pulse. The transmitted long pulses can be modulated to increase their bandwidth, and on receive these pulses are compressed in the time domain, resulting in a range resolution higher than that associated with a non-modulated pulse. However longer pulses increase the blind range simply because the receiver has to be gated off during transmission. One way to recover the near range is to transmit a separate short 1-4244-1539-X/08/$25.00 ©2008 IEEE pulse after the long pulse has been fully received. The echoes from the long pulse are from far range targets and echoes from the short pulse are from targets at near ranges. This method is simple but inefficient in that the transmission and reception of the short pulse unnecessarily increases the duration between two long pulses. Consequently the radar receives fewer pulses from far range targets within the 3 dB beamwidth. An alternative more efficient method is to form a complex pulse with both long and short pulses concatenated together. After transmission of all sub-pulses, the receiver collects reflected signals from all subpulses and separates them based on either their frequency or modulation. Although the concatenated waveforms have a number of attractive attributes, they inherit a challenging problem in that different sub-pulses will reflect back from different ranges but be received at the same time. Since each sub-pulse experiences different attenuation, it therefore requires its own sensitivity time control (STC) [2]. However at the RF front end these subpulses are not separable and a traditional STC scheme can not be directly applied. There is a requirement to configure a more appropriate STC. In this paper we present a system for concatenated waveform and discuss its limitations. We also propose a feasible STC scheme to address the STC requirements for concatenated waveforms. This paper is outlined as follows. The system and method for concatenating pulses are presented in Section II. A detailed analysis of the STC requirements for the complex waveforms is presented in Section III and a feasible STC approach is proposed in Section IV. A sample analysis is presented in section V and summary is presented in Section VI. II. METHOD AND SYSTEM FOR CONCATENATED WAVEFORMS A modulated pulse with a long pulse length can be found in many radar systems. For monostatic radar, the receiver has to be gated off during transmission, resulting in a minimum blind range of: Rb = cτ L 2 (1) where c is the speed of light, and τL is the uncompressed pulse length. A separate short pulse, often referred as a “fill pulse”, can be transmitted to cover this blind zone as shown in Fig. 1 (a). However the dwelling time on the far range targets is reduced and the corresponding loss is approximately: Lb = 1 − τ L +τ S TR + 2τ L + τ S (2) TR is the maximum receiving time determined by the maximum unambiguous range, and τS is the short pulse length which is in general the same as the compressed long pulse. In air surveillance radar, the ratio between the total pulse length (long + short) and the maximum receiving time is around 0.1 and the corresponding loss is about 0.5 dB according to Eq.(2). In other words, 10% of the transmitter’s power is wasted. Long Short Long Short Transmit 100 1 TR Receive Long Short (a) Transmit 101 Receive Short Long (b) Fig. 1 – Example pulse schemes for simultaneous far and near range coverage: a) separate long and short pulses b) concatenated long and short pulses. The lost transmitter power can be saved by using concatenated waveforms such as a simple example shown in Fig. 1(b). In this example a short pulse is transmitted immediately following the long pulse. Both short pulse and long pulse echoes are received simultaneously but from different ranges. The short pulse returns are separated from long pulse returns based on their different characteristics. The data can be processed independently or realigned to form the full range coverage for further common processing. Different features can be utilized to distinguish the echo signals from each pulse, which includes orthogonal coding, polarization or frequency. The concatenated pulse waveforms described herein generally consist of sub-pulses with a certain order within a given pulse repetition interval (PRI). Pulses are arranged such that longer pulses are transmitted first followed by shorter pulses to provide a fill for a blanking interval. Therefore, if the required minimum detection range is within the blind zone of long pulse (LP), the concatenated waveforms include a short pulse (SP) at the end to achieve the minimum range. Since the SP already has good range resolution, pulse compression may not be necessary. Another attractive attribute of concatenated waveforms is their flexibility to form complex waveforms for enhanced processing gain. For example, frequency diversity can be achieved within one PRI with only one transmitter required. An example of this waveform is shown in Fig. 2, which is specifically applied to an L band En Route radar. (Details can be found in [4]). This waveform has better azimuth resolution and accuracy compared to waveforms having PRI alternative frequency diversity. Fig. 2 – Example of concatenated waveforms with frequency diversity. In Fig. 2, carrier frequency offsets are introduced for each sub-pulse which simplify the implementation and also increase the pulse compression freedom. Any modulation can be applied to one of the pulses regardless of the modulation applied to the other pulses. Specific numbers have been shown only for example with the understanding that other values can be selected for the IF and RF frequencies of these pulses as well as different time durations. These values are dictated by the specific application where a unique requirement may exist for minimum range, maximum range, and range resolution etc. In this case, there are two long pulses optimized for target detection over the far range, and there are two medium pulses (MP) optimized for near range target detection. The minimum coverage requirement for En Route radar is relaxed and there is no need for a short pulse. The frequency offset between the long pulses is set to provide sufficient de-correlation for frequency diversity and separation for filtering. The frequency offset between long and medium pulses is only constrained by the filtering requirement. In Fig. 2 one set of LP and MP is centred at carrier frequency F1 with +/- 3 MHz; and another set is centered at F2 with +/- 3 MHz. It should be noted that the pulses do not have to be centered about the carrier frequency, although this simplifies the generation of these pulses and the processing of subsequent reflections. A sample portion of a radar receiver is shown in Fig. 3 for processing reflections of the concatenated pulse waveform of Fig. 2. After a low-noise amplifier (LNA), the return signals are separated into F1 or F2 based frequency paths through two bandpass filters. The F1 centered filter has a sufficient bandwidth to pass both F1 LP and F1 MP reflected signals. A similar characteristic applies to the F2 centered bandpass filter. The mixers then down-convert the output of the bandpass filters respectively to the IF band based on oscillation frequencies LO1 and LO2. The reflected MP waveform is then separated from the reflected LP waveform by a second stage bandpass filter. The separated signals are digitized and pulse compressed. III. STC ANALYSIS Sensitivity time control is in general required in the RF front end to protect the receiver from saturation due to nearby clutter. STC reduces the receiver gain at close ranges where clutter echoes are large and maximum sensitivity is not generally required. The gain will continually increase as echoes return from further range and reach the maximum when the clutter barely exceeds the noise level. In air traffic control radars STC is applied to control land clutter as well as angel clutter (such as birds and insects). The selection of gain versus time characteristic is dictated by the type of clutter encountered. The use of concatenated waveforms makes the application of STC more complex since different pulses reflect back at the same time but from different ranges. Ideally a distinct STC value should be selected for each pulse depending on its true range. However this is not practical since it is not feasible to separate the pulses at the RF front end. It can be noted that a similar but more extreme result exists in high PRF pulse Doppler radar [1], where STC is not feasible due to the excessive number of multi-trip signals. where γ is the parameter describing the surface property and ψ(R) is the grazing angle. For range R the grazing angle can be determined by [6]: h h 1 + 2r R e ψ ( R ) = sin −1 R − 2r e ≈ sin −1 h − R R 2r e (8) where h is the effective radar height (site elevation above mean sea level plus radar antenna height minus terrain elevation above mean sea level), R is the radar slant range and re is the effective earth’s radius. Although for low grazing angle this model tends to overestimate the reflectivity due to propagation factors, it still provides a basis for the analysis with a loose upper bound of the clutter’s power as a worst-case scenario. A high grazing angle of near 90° is not of interest since our primary focus is ground based radar with grazing angle generally less than 45° . At range R the land clutter will arrive at the radar receiver with the following radar cross sections (RCS): σ ( R) = σ ° (R ) A(R ) = γ tan(ψ ( R ))Rθ B cτ 2 (9) where θB is the 3-dB antenna beamwidth, A(R) is the radar resolution cell and τ is the compressed pulse length (assume all pulses have the same length after pulse compression). In the following analysis we assume tan(ψ) to be 1, which provides more conservative clutter power estimation for grazing angles less than 45° . At time instant t, the clutter power from the first long pulse is as follows on the receiver front end: Fig. 3 – Example of receiver processing of the 4 – pulses waveforms. In this section the STC requirements are analyzed for the waveform with specific parameters as shown in Fig. 2. However the analysis procedures and conclusions can be readily extended to other concatenated waveforms. At time instant t the received signals are reflected from different ranges for different pulses and determined as follows: RL1 (t ) = (t + T + τ M 2 + τ M 1 + τ L 2 + τ L1 + 3∆ ) c / 2 (3) RL 2 (t ) = (t + T + τ M 2 + τ M 1 + τ L 2 + 2∆ ) c / 2 (4) RM 1 (t ) = (t + T + τ M 2 + τ M 1 + ∆ ) c / 2 (5) RM 2 (t ) = (t + T + τ M 2 ) c / 2 (6) where τL1 is F1 (first) long pulse length, τL2 is F2 (second) long pulse length, τM1 is F2 (first) medium pulse length, τM2 is F1 (second) medium pulse length, t is relative time to receiver start-up, ∆ is gap among pulses for spectrum control, T is receiver start-up time after complete of transmission and c is the speed of light. RL1, RL2, RM1, RM2, are range of reflected signals for first LP, second LP, first MP and second MP, respectively. For land clutter backscattering coefficient σ0, we adopt the constant-γ model which is in excellent agreement with most measurements at grazing angles not too close to 90° or 0° [5], and the model for range R is: σ °(R ) = γ sinψ (R ) (7) cτ 2 λ 2 C L1 (t ) = 4 i =0 (4π )3 RL1(t ) − i cτ 2 cτ 2 N L1 −1 Pt Gt Gr γθ B λ 2 = 3 i =0 (4π )3 RL1 (t ) − i cτ 2 N L1 −1 Pt Gt Grσ RL1 (t ) − i ∑ (10) ∑ where Pt is peak transmitted power, Gt and Gr are respectively the transmit and receive antenna gains, λ is radar wavelength and NL1 is the pulse compression ratio for the first long pulse. It is worth noting that integration over the uncompressed pulse length is important for the estimation of land clutter strength. This integration, often neglected, shows the correlation between different range cells in order to provide this land clutter strength estimate. In a similar way we can obtain the clutter power received from the other pulses: cτ 2 N L 2 −1 Pt Gt Gr γθ B λ 2 C L 2 (t ) = (11) 3 cτ 3 i =0 (4π ) RL 2 (t ) − i 2 ∑ N M 1 −1 C M 1 (t ) = ∑ i =0 cτ 2 λ 2 3 (4π )3 RM 1 (t ) − i cτ 2 Pt Gt Gr γθ B (12) cτ 2 λ 2 C M 2 (t ) = (13) 3 cτ 3 i =0 (4π ) RM 2 (t ) − i 2 At a given receiver time t, STC may be required due to the following reason(s): 1) To prevent the land clutter power from saturating the receiver. The receiver saturation happens when the maximum signal power received Psa causes a LNA gain compression of one dB. STC has to be applied at the RF front end. The STC required for this purpose is*: N M 2 −1 ∑ Pt Gt Gr γθ B STC LLC 1 (t ) = C L1 (t ) − Psa + δ STC LLC 2 (t ) = C L 2 (t ) − Psa + δ STC MLC1 (t ) = CM 1 (t ) − Psa + δ (14) where δ is the margin to protect the LNA from saturation. 2) To prevent A/D converter saturation, STC can be applied at any point prior to the A/D converter. This is because in general the A/D converter has a smaller dynamic range than the analog receiver and sets more aggressive requirements on STC. The STC required for this purpose is: STC LAD 1 (t ) = C L1 (t ) − N fl − Dr + δ n N fl − Dr + δ n STC MAD1 (t ) = CM 1 (t ) − N fl − Dr + δ n STC can be applied either at the RF stage, IF stage and/or digital section of the system. Due to the complexity of concatenated waveforms, a combination of these STCs is proposed. The proposed scheme works ideally for the situation where the most demanding requirement is for angel clutter control, and least demanding requirement is for amplifier saturation. If this situation does not exist, the designer has to be aware of the possible integration loss and diversity loss for close in targets. This might not be an issue for low flying targets, but might be a concern for close-in height coverage especially with cosecantsquared beampattern, such as air traffic control radars. A STC MLC2 (t ) = CM 2 (t ) − Psa + δ STC LAD 2 (t ) = C L 2 (t ) − IV. STC SCHEMES FOR CONCATENATED WAVEFORMS At the RF front end, RF STC is required to prevent amplifier saturation. At this stage, all pulses are non-separable, and only one STC can be applied. The STC is determined as: STC RF (t ) = ( LC LC LC max STC LLC 1 (t ), STC L 2 (t ), STC M 1 (t ) , STC M 2 (t ), 0 STC LBC 2 (t ) = S L 2 (t ) − N fl − Vx STC MBC1 (t ) = S M 1 (t ) − N fl − Vx IF STC ( ( ( ( RF STC LIF1 (t ) = max STC LAD (t ), 0 1 (t ) − STC STC LBC 1 (t ) = S L1 (t ) − N fl − Vx (16) STC MBC2 (t ) = S M 2 (t ) − N fl − Vx where the Vχ is visibility factor which can be determined based on the system requirements for false alarm rate and probability of detection for the largest birds of interest. SL1(t), SL2(t), SM1(t), and SM2(t), are respectively the received power from the largest birds of interest at different pulses. More than often, receiver (LNA) requires the least STC attenuation, and angel clutter requires the most STC attenuation. This is at lease the case in the later example. (17) The IF STC should be applied prior to A/D converter. At this stage data from different pulses have been separated. Individual STC can be applied. The required IF STC is as follows: STC MAD2 (t ) = CM 2 (t ) − N fl − Dr + δ n where Nfl is the system noise floor, δn is the margin required to protect A/D converter from clipping, and Dr is the A/D converter dynamic range. 3) To prevent angel clutter (birds) from saturating radar. This is especially required for air traffic control radars. STC can be applied in analog or in digital sections. The required STC is: ) As shown in Equation (17), amplifier saturation is avoided by selection of the largest STCRF(t) value over the long and medium pulses. B (15) RF STC ) ) ) ) RF STC LIF2 (t ) = max STC LAD (t ), 0 2 (t ) − STC IF AD RF STC M 1 (t ) = max STC M 1 (t ) − STC (t ), 0 STC MIF2 (t ) = max STC MAD2 (t ) − STC RF (t ), 0 (18) The IF STC can be applied anywhere between the data separation and A/D converter. However special consideration has to be given if Constant False Alarm Rate (CFAR) techniques are adopted in the later detection stage in that the IF STC can alter the noise floor. The combination of RF STC and IF STC helps to prevent the A/D converter from clipping due to any of the pulses. When RF STC sets a higher attenuation value, there will be no need to have an IF STC. C Digital STC Individual digital STC can be applied anywhere after A/D converter. The STC required is as follows: RF STC LD1 (t ) = STC LBC (t ) − STC LIF1 (t ) 1 (t ) − STC RF STC LD2 (t ) = STC LBC (t ) − STC LIF2 (t ) 2 (t ) − STC STC MD 1 (t ) = STC MBC1 (t ) − STC RF (t ) − STC MIF1 (t ) (19) STC MD 2 (t ) = STC MBC2 (t ) − STC RF (t ) − STC MIF2 (t ) * In this paper, the positive STC value means attenuation; zero and negative STC values means no STC required. All units are in dB. The digital STC can bring artefacts to the noise floor in a similar way to that of the IF STC. When CFAR is employed, the threshold has to be carefully designed to obtain a real constant false alarm rate. A feasible design is to apply a minimum threshold map during the detection process, which will regulate the detection of the smallest target of interest under the consideration of combined IF and digital STC effects. The application of digital STC is flexible in that it can be combined and applied together with IF STC. Another way to apply digital STC is to incorporate it into the detection process instead of physically applying the attenuation, which avoids the side effects on noise floor. When the combination of RF STC and IF STC sets a higher attenuation value, there is no requirement for digital STC. For the four sub-pulses concatenated waveform or similar waveform which requires integrating the sub-pulses, it is recommended to apply the digital STC prior to any video or binary integrator. The combination of RF, IF and digital STC helps to control the angel clutter breakthrough from any of the pulses. The other curves in Fig. 5 represent the STC requirements for angel clutter control with a maximum RCS of 0.01m2. If the STC required for the second medium pulse was applied at the RF stage, then we would be applying the correct STC for the second medium pulse but too much for the other three pulses. For example at receiver time 1µs, we would have the correct STC for the second medium pulse but 35 dB too much for the echo from the second long pulse which is coming back from a range of 13.77 nmi at the same time. This would result in the sensitivity loss, which can not be made up for later in the system. To circumvent this problem we should defer the required STC for angel clutter control to digital stage, but only apply the RF STC determined by the amplifier saturation. STCs Required for the Four Pulse Waveform In this section the STC is designed for the waveform shown in Fig. 4 with parameters that are typical for modern L band air surveillance radars. Both long pulses have a length of 120µs and effective length of 110µs with tapering at the start and end of each pulse. Both medium pulses have a length of 22µs and effective length of 20µs with tapering. The gap in any two neighbouring pulses is 5µs. All pulses will be compressed into 1µs. The transmitter peak power is 50 kW and the carrier frequency is 1.3 GHz. The antenna has a gain of 34 dB. A typical limit level for the receiver protector is -10 dBm and a 10 dB margin has been added during the calculation. The receiver is gated on at 1µs after the completion of transmission. The instrumental coverage is from 5 nmi to 200 nmi. Land clutter backscattering coefficient σ0 is set to -10 dB (m2/m2 ) which is a severe clutter scenario [5] for a rural environment. The analysis has shown that at the RF stage the second medium pulse requires the most attenuation and determines the RF STC according to Eq. (17). The RF STC is shown in Fig. 5 as the star-labeled curve. The abscissa indicates the range of the second medium pulse’s returns at receiver time t. For example, at receiver time 1µs, the radar returns of second medium pulse is from 1.78 nmi, while the returns of first medium pulse, second long pulse and first long pulse are actually from 3.97 nmi, 13.77 nmi and 23.89 nmi, respectively. Fig. 4 – Example of the 4-pulses waveforms with specific timing and reflection range. STC (dB) V. EXAMPLE ANALYSIS 85 80 75 70 65 60 55 50 45 40 35 30 25 20 15 10 5 0 -25 -20 -15 -10 -5 0 2nd Med Pls 1st Med Pls 2nd Lng Pls 1st Lng Pls RF STC for all 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Range (nmi) Fig. 5 – RF STC and the individual STCs required by angel clutter control There is no need for the IF STC in that the A/D converter has 16 bits which provides sufficient dynamic range. The digital STC is then determined by the difference of the applied RF STC and the desired STC for angel clutter control according to Eq. (19). The digital STC curves are presented in Fig. 6 for each pulse, respectively. Except for the early receiver time (before 4µs), the RF STC is less demanding than all the other angel clutter STCs. Therefore there is an optimal solution for most of the receiver time (after 4µs), which is to apply RF STC at the RF front end and apply the difference in as digital STCs according to Eq. (19). VI. SUMMARY In this paper we analyze and propose a multi-stage STC scheme for complex concatenated waveforms. The principle is to apply minimum STC at the RF front end to satisfy the receiver saturation protection from close in clutter. Additional pulse-specific STCs are applied at the IF and/or digital stages where the pulses have been split and extra flexibility exists. When the STC required by receiver saturation is least demanding, our proposed scheme can provide an optimal solution as if all pulses could be separated at the RF front end. If this is not the case, a loss in sensitivity is unavoidable. Digital STCs Required for the Four Pulse Waveform 45 40 35 STC (dB) 30 2nd Med Pls 1st Med Pls 2nd Lng Pls 1st Lng Pls 25 20 15 10 5 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Range (nmi) Fig. 6 – Digital STCs for each of the pulses ACKNOWLEDGMENT The first author would like to thank Dr. A. M. Ponsford for his valuable comments. REFERENCES [1] M. I. Skolnik, Introduction to Radar Systems. Third Edition, New York, McGraw-Hill Press, 2001. [2] M. I. Skolnik, Radar Handbook. New York, McGraw-Hill Press, 1990. [3] A. W. Rihaczek, Principles of High-Resolution Radar. Peninsula Publishing, 1985. [4] T. M. Chan, M. Gerecke, “Method and System for Concatenation of Radar Pulses”, US Patent Application # 11/832,973 filed August 2, 2007. [5] D. Barton, Radar System Analysis and Modeling, Norwood, US, Artech House, 2005. [6] F. E. Nathanson, Radar Design Principles, Second Edition, New York, McGraw-Hill Press, 1991.