A Low-Complexity Noncoherent IR-UWB Transceiver ... - IEEE Xplore

IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 54, NO. 4, APRIL 2006

1637

A Low-Complexity Noncoherent IR-UWB Transceiver Architecture With TOA Estimation Lucian Stoica, Student Member, IEEE, Alberto Rabbachin, Student Member, IEEE, and Ian Oppermann, Senior Member, IEEE Invited Paper

Abstract—Impulse-radio (IR)-based ultra-wideband (UWB) technology is a strong candidate for short-range data communication and positioning systems. This paper examines the performance of time-of-arrival (TOA) position estimation techniques as well as the simulated and measured performances of an IR-UWB noncoherent energy-collection receiver. The noncoherent IR-UWB transceiver has been designed for operation over the frequency range 3.1–4.1 GHz and implemented in 0.35- m SiGe BiCMOS technology. The performance of two different algorithms, namely, the threshold-crossing and the maximum selection (MAX) algorithms, are compared in terms of TOA estimation error in Saleh Valenzuela channel model 3 and channel model 4. The implemented structure of the TOA MAX algorithm suitable for IR-UWB-based noncoherent receivers is presented. A UWB testbed has been constructed in order to test and measure the transmitted waveform as well the receiver performances. The simulated receiver noise figure is 7.3 dB while the receiver gain is 34 dB. The TOA MAX algorithm can achieve 5-ns positioning accuracy for 95% of cases. Constant transconductance tuning circuits for improved TOA estimation reliability are also presented. Index Terms—Impulse-radio ultra-wideband (IR-UWB), noncoherent low-complexity transceiver architecture, RF front-end, SiGe BiCMOS, tag networks, time-of-arrival (TOA) estimation.

I. INTRODUCTION LTRA-WIDEBAND (UWB) has grown in popularity in the years since the Federal Communications Commision (FCC) regulations in the United States [1] have driven the needs of consumer and military applications and developments in solid-state technology and communications. Impulse-radio (IR)-based UWB (IT-UWB) technology utilizes signals with very low spectral densities, is resistant to channel multipath, has very good time-domain resolution allowing for location and tracking applications, and is relatively low-comlexity and low-cost. Lately, IR-UWB technology has been used in low-cost RF sensor or “tag” networks [2], [3]. With its inherently high time-of-arrival (TOA) estimation accuracy, UWB enables tags to be tracked and located with high precision within a specified area. IR-UWB tags are able to transmit and receive signals at extremely low power levels and are targeted at indoor networks with data rates of up to 100 Kb/s. The functionality of the

U

Manuscript received August 1, 2005; revised January 17, 2006. This work was supported in part under the PULSERS Project, by The National Technology Agency of Finland (Tekes), by the Nokia Foundation, by the HPY:n Foundation, and by Infotech Oulu. The authors are with the Centre for Wireless Communications, University of Oulu, Oulu FI-90014, Finland (e-mail: [email protected]). Digital Object Identifier 10.1109/TMTT.2006.872056

noncoherent energy-collection receiver is based on recovering the signal energy spread in the UWB channel. This leads to a fundamental tradeoff in designing IR-UWB energy-collection tags: there is no need for a channel estimation block with the drawback of interference and noise enhancement [3], [4]. When applying conventional coherent structures for UWB receivers, the optimal exploitation of the pulse timing accuracy is only possible with extremely high-frequency clocks on the order of tens of gigahertz, which are capable of sampling subnanosecond time windows. A noncoherent energy-collection design approach to IR-UWB tags has been presented in [3]–[5]. In the system considered, the mobile UWB tags operate under the control of fixed centralized nodes at data rates of a few kilobits per second. The communication between the fixed nodes and mobile tags takes place in a time-division duplex mode with an access point in bursts of 100 s once every 100 ms, which gives a duty cycle of the order of 0.1% [2], [3]. What makes IR-UWB a leader candidate technology for indoor positioning systems is the fine time resolution associated with the short-time-duration pulses. In general, positioning techniques exploit one or more characteristics of the radio signals to estimate the position of their sources. One of the traditional positioning techniques is TOA [6], [7]. IR-UWB noncoherent receivers can implement TOA positioning techniques using a bank of overlapping or nonoverlapping integrators. By integrating the received signal in small time windows over a symbol period and then selecting the integrator which gives the maximum value [8], a coarse TOA estimate can be produced. The integration windows must be sufficiently small to satisfy the desired uncertainty of the TOA estimate. Smaller integration windows also support greater consistency of the integrated values across process, voltage, and temperature (PVT) variations. The control of the integration values within 1% accuracy can be achieved by using a tuning circuitry [9]. In this paper, we propose an IR-UWB noncoherent receiver with TOA capabilities which makes use of current-controlled constant transconductance tuning circuitry. A number of authors have studied TOA estimation for IR-UWB signals. In [10], correlation techniques in conjunction with aserial search have been proposed for TOA estimation. In [11], special code design has been considered as a means to facilitate the estimation of TOA with the correlation approach. Frequency-domain approaches and chip-level postdetection integration (CLPDI) have been considered in [12]–[14]. Recently, a number of authors have studied integrated noncoherent IR-UWB transceiver architectures together with TOA positioning.

0018-9480/$20.00 © 2006 IEEE

1638


An FCC-certified “UWB Precision Asset Location” transceiver which is able to achieve position accuracies of better than 30 cm, and operates in the -band region of the spectrum, has been described in [15]. In [15], a reference tag is used to fully calibrate the system, and the waveform used is a 400-MHzbandwidth UWB pulse. A discrete hardware transceiver architecture operating in VHF/UHF bands which utilizes a singlepulse UWB detection capability as well as discrete transceivers operating in the -band were reported in [16]. Compared with discrete hardware transceiver architectures published in [16], we present the integrated version of a low-cost and low-complexity noncoherent energy-based IR-UWB transceiver architecture appropriate for a low-cost tag concept. Also, the performances of two TOA estimation algorithms suitable for noncoherent energy-based IR-UWB receivers are presented. In [17], a 1.2-V 6.7-mW IR-UWB baseband transceiver architecture was been presented. A 0.13- m CMOS low-noise amplifier (LNA) including a third-order Chebyshev bandpass-matching network based on the PCSNIM [18] concept was presented in [19]. The LNA presented in [19] has a 3-dB bandwidth of 7.7 GHz with an input return loss better than 10 dB and an output return loss better than 15 dB over the entire bandwidth. Another systematic approach of a 3–5-GHz UWB CMOS LNA together with two output impedance-matching methods was presented - and -parameters are in [20]. In this paper, both the below 10 dB, the overall gain is 13.5 dB, and the noise figure (NF) is below 2 dB. The system bandwidth, gain, and NF of our RF front-end are 3.1–4.1 GHz, 34 dB, and 7.3 dB, respectively. The integration technology of the IR-UWB transceiver is 0.35- m SiGe BiCMOS. The remainder of the paper is organized as follows. In Section II, the network architecture is presented. In Section III, we present two TOA algorithms: the threshold-crossing (TC) and maximum selection (MAX) algorithms, as well the implementation structure of the MAX algorithm into IR-UWB noncoherent receivers. The IR-UWB noncoherent transceiver architecture is presented in Section IV. In Section V, we present a performance comparison between the TC and MAX algorithms, the measured waveform of the transmitted UWB pulse, the receiver output spectrum, as well as the IR-UWB testbed. Concluding remarks are given in Section VI. II. SYSTEM CONCEPT Low-data-rate tag networks with location and tracking capabilities are an increasingly popular application for UWB technology [2], [3]. The central system utilizes the information coming from the fixed nodes (FNs), which are closely time synchronized by sharing the same local clock through cable connections. The FNs are positioned at known coordinates in the area being monitored, and we assumed that they are perfectly synchronized. The multiple-access interference (MAI) is minimal, since each tag transmits data in different preassigned time slots. Synchronization between the FNs and the mobile tags is performed once every second due to the drift in the clock of the mobile tags as well as the FNs. This is achieved by broadcasting a beacon from one of the FNs. The TOA of the beacon is used as the reference clock for the mobile tags to transmit data according to the preassigned time slots. For

3-D positioning, four FNs are needed to obtain exact solutions using TOA measurements. The positions of the tags are to be estimated. For an overdetermined system, several different approaches have been proposed, such as spherical interpolation [6], [7], [21], [22], the two-stage maximum-likelihood method [23], and the linear-correction least-square approach [24]. We propose a two-stage approach for fast timing acquisition in order to obtain the TOA of the desired signal. The received signal is amplified by the LNA and variable-gain amplifier (VGA) and then squared. The first stage is a coarse synchronization which is implemented to obtain the estimated position or area of the energy clusters of the received signal without knowing the position of the peak of the particular cluster. The next step is a fine synchronization where the objective is to locate the peak energy and collect the energy for that integration window. The fine synchronization stage can be used for ranging. The fine synchronization is done with the same set of integrators that are used on the coarse synchronization process. This will therefore reduce the hardware complexity of the tag. The main differences between ranging and fine synchronization are as following: the ranging requires knowledge of the first energy cluster which is assumed to contain the first path required for delay estimation, while synchronization requires knowledge of as many clusters as possible since maximum energy collection is used. The fine synchronization stage is impleintegration windows within the coarsely mented by placing synchronization windows. In this way, the searching process of the starting point of the cluster will be more refined. If the firststage search is successful, the coarse TOA estimate will satisfy

(1) where is the estimated delay, is the optimal estimated delay, is the integration time interval. and III. ENERGY-COLLECTION-BASED TOA ESTIMATION Positioning techniques exploit one or more characteristics of radio signals to estimate the position of their sources. Some of the parameters that have been traditionally used for positioning are the received signal strength intensity (RSSI), the angle of arrival (AOA), and TOA. The estimation of AOA, on the other hand, requires multiple antennas (or at least an antenna capable of beam-forming) at the receiver. This requirement implies size and complexity requirements that are often not compatible with the low-cost, small-size constraints associated with applications such as wireless tag networks for which UWB technology is particularly suited. Therefore, TOA stands out as the most suitable signal parameter to be used for positioning with UWB devices. However, due to the ultrashort (usually subnanosecond) pulses, it poses challenges for synchronization in UWB systems. Some techniques have been proposed to estimate the TOA of UWB signals, for instance, correlation in conjunction with serial search [10], special code design [11], and frequency-domain processing [12]. However, all of the above solutions seem to be in conflict with the strict requirements of low cost and low complexity imposed on some UWB applications and may not provide satisfactory TOA estimates. Another TOA-estimation

STOICA et al.: LOW-COMPLEXITY NONCOHERENT IR-UWB TRANSCEIVER ARCHITECTURE WITH TOA ESTIMATION

1639

Fig. 1. TOA estimation.

scheme for UWB signals is the generalized-likelihood ratio test [25]. However, this technique is relatively high in complexity. In order to further reduce the complexity of UWB systems, noncoherent receivers using energy-collection [26] and transmitted-reference approaches [27]–[29] have recently been proposed. We will first provide a detailed description of the energy-collection-based approach. A two-stage TOA estimation scheme will then be presented. The block diagram of the energy detection receiver is presented in [3] and references therein. The drawback of the noncoherent approach is noise enhancement due to the squaring and the degradation in time resolution that is proportional to the length of the integration. As a consequence of the energy-collection approach of the receiver, a low NF of the receiver is highly desirable. TOA estimation can also be performed using energy-detection structures such as the noncoherent IR-UWB receiver architecture presented in [3]. In the first stage, a bank of integrators is employed. Each integrator integrates the squared symbol for of one symbol period , as shown in Fig. 1. a fraction A search is performed over one symbol duration. The first integrator starts integration at a chosen time point. Each of the comother integrators begins integration after a delay of pared to its preceding integrator. The start time point of the integrator whose output is the maximum among all of the integrators provides a coarse TOA estimate. The difference between the TOA estimate and the chosen starting point of the first inte. With a probability that grator is denoted as the time error is dependent on the SNR, the coarse TOA estimate will indicate the region containing the first received pulses. After initial synchronization is completed, the TOA estimation is performed around the synchronizaby dividing the uncertainty region tion point into integration windows, where represents the number of integrators available in the receiver. Intuitively, the estimation accuracy is dependent on the uncertainty region size and on the number of integrators. As opposed to symbol synchronization, which provides the time reference that ensures the maximum signal energy detection, the TOA estimation can be seen as fine synchronization, searching for the arrival time of

Fig. 2. MAX algorithm implementation structure in IR-UWB noncoherent receivers.

the received signal. Based on the energy measurements, a decision is made according to a chosen criterion. For example, a TC criterion can be used. With TC, the search is performed serially and is stopped once a measurement value crosses the threshold. The corresponding window is then chosen, and its starting point provides the TOA information. If necessary, a verification process may be pursued. For example, new measurements are taken from the chosen window and are tested against the threshold. If the threshold is crossed out of tests, the chosen window is finally accepted. Otherwise, the search resumes. In the event that no measurement crosses the threshold, new measurements are taken and the search resumes. The TC algorithm requires the setting of a threshold. The other approach is the MAX criterion. With this criterion, measurements at all windows are first compared. Then, the maximal measurement is produced, and the relevant window is selected. In the event that no appropriate thresholds can be readily obtained, the MAX criterion could be desirable. Another criterion is the hybrid of MAX and TC. In this hybrid criterion, the maximal measurement is first obtained. Then, the maximum is examined against the threshold. If the threshold is crossed, the related integration window is selected. If the maximum does not cross the threshold, the search resumes. The TOA MAX algorithm implementation structure of the noncoherent receiver is presented in Fig. 2. In the following, we will describe the basic functionality of the TOA implementation structure. First, the capacitor C is reset (R is active), and then the output of first integrator (INTEGRATOR1) is fed into the comparator. If the INTEGRATOR1 output is bigger than the

1640


is the th chip of the pseudorandom (PR) code, where is the transmitted pulse with pulsewidth ps, is the with is the chip interval, and symbol interval, ns is the delay used to distinguish different transmit . The received signal after the receiver ansymbols tenna is given by

Fig. 3. Current-controlled constant transconductance tuning for the MAX algorithm implementation structure in an IR-UWB noncoherent receivers.

(3) voltage across C, then the capacitor C will be loaded (LOAD is active) with the output voltage of INTEGRATOR1, otherwise it does not change value (LOAD inactive). The same decision process as described above is applied to the rest of the integra. The integrator position tors INTEGRATORSi, which gives the maximum value among all of the integrators will be sent to the baseband as the TOA estimate. The timing in Fig. 2 denote the integration windows signals , , and integrators. The minimum integration window is of all 5 ns, which imposes the requirements over the gain, power consumption, sensitivity, and dynamic range of the integrators. Specifically, the transconductor gain must be sufficiently high to allow the signal at the output to be within the resolution and range of the following comparator. The sensitivity has to be such that the minimum signal of their inputs is still integrated to a useful value. The signal is converted to baseband after the squaring. We use one amplifier stage with digitally controllable variable gain after the squaring, such that the signal is at the level required by the integrators. Since we are using a integrator filter, we can expect a 30% tolerance on the absolute ratio [9]. Before synchronization and TOA value for the integration, the integrators needs to be tuned off-chip using a schematic presented in Fig. 3 such that the tolerance of is reduced to 1% [9]. The goal is to set the transconductance to the inverse of an external resistance . The cirvalue is smaller cuit presented in Fig. 3 works as follows: when (bigger) than the reference value , then the voltage at is less (higher) than and the differential the top of . At equilibrium pair will increase (decrease) the value of state, the differential voltage at the input of the differential pair will equal . We mention that, even will be zero, and if the transconductor structure we are using in the integrator is differential, we can still make use of the single-ended output element, as shown in Fig. 3. In Fig. 3, denotes of the all of the parallel elements as shown in [3, Fig. 6], such that the tuning is performed over all integrators. IV. TRANSCEIVER ARCHITECTURE The architecture for the IR-UWB tag transceiver as well as details of modulation type are presented in [3]. The transmitted signal for one user of interest is given by

(2)

where is the number of resolvable paths, defines the gain is the first derivative of , and is for path , zero-mean additive Gaussian noise. The large number of distinguishable multipath components, which are available at the receiver due to the large bandwidth of the transmitted IR-UWB signal [30], will offer a gain in terms of both positioning accuracy and diversity. Due to hardware complexity limitations, the IR-UWB tag design is targeted to have a very simple structure, thus leaving all of the positioning computational capability to the fixed network. A. Transmitter Architecture The transmitter module contains a 533-MHz clock generator, a UWB Gaussian pulse generator, and a UWB dipole antenna [3]. The UWB pulse generator produces the Gaussian monocycle of 750 ps, which is the typical width. The pulse generators offer the possibility of selecting the pulse repetition frequency to be the the on-chip generated 533-MHz frequency or any off-chip frequency. The transmitter’s -parameter and transmitted pulse waveform are presented in Figs. 8 and 9, respectively. B. Front-End Receiver Design The front-end receiver on chip consists of an LNA, VGA, and Gilbert mixer [31]. A general guideline for the design of each block is to have a total NF of the receiver chain below 10 dB, low power consumption, and the ability to deal with the high-frequency signal with some gain. The impedance match to 50 is included in the LNA design. The input of the LNA is differential, as is the rest of the front-end receiver until the integrator. The differential signal is obtained by feeding the signal through an RF-balun after filtering. The RF-balun is also an OFF-chip component. 1) LNA: The LNA design used in the receiver has a fully differential structure which was converted from its single-ended equivalent by the mirror principle. The schematic is depicted in Fig. 4. The values of the LNA components presented in Fig. 4 nH, nH, nH, are as follows: . The values of the input and output capacand itors are 2 pF. This kind of a structure is also presented in [32] and [33]. The input impedance is matched to 50 so that each input is matched to 100 . The impedance match is realized by a wideband LC-ladder bandpass filter consisting of inductors and in conjunction with the capacitors of the input tranand a parasitic capacitors of the input pad and sistors ON-chip inductor . The input transistors are dimensioned to


1641

Fig. 5. Schematic diagram of the UWB VGA.

Fig. 4. Schematic diagram of the UWB LNA.

be large in order to minimize the base resistance , which is a considerable source of noise. The NF of the LNA for a given resonance frequency is determined by [32]

NF

(4)

where NF is the noise figure, is the base resistance of the input transistor , is the 50- source resistance, is , and is the unity gain frethe transconductance of . The cascode transistors and improve the quency of output–input isolation and decrease the Miller effect. The bias and . The load of the amplifier is set by the bias voltages and inductor . consists of resistor 2) VGA: The VGA design presented in Fig. 5 is based on the Gilbert cell. This approach is presented in numerous papers, to mention a few [34]–[36]. The VGA gain is adjusted by altering . The larger the voltage is, the more the control voltage and and hence the current is directed through transistors and is decreased, current drawn by the gain transistors which reduces the gain of the amplifier. The NF has a minimum when the maximum gain is used. This is due to the fact that and are switched off and contribute no the transistors noise when all of the current is directed through transistors and . The maximum noise occurs with a 6-dB gain reduction when all of the quad transistors draw equal currents [36]. The and . input gain stage of the VGA consists of transistors and The bias of the VGA is set by the bias voltages

through resistor , which isolates the bases from each other. is used to improve linearity and The degeneration resistor improves the match between the two current sinks used as the bias current of the circuit. 3) Gilbert Multiplier: The squaring circuit of the receiver is a Gilbert multiplier. The output of the VGA is buffered by two different output stages to exclusively drive both input ports of the Gilbert cell. Since the input impedance of the two input ports are slightly different, the multiplier suffers from a small phase error. However, this is of no big importance in our noncoherent system. As the squaring operation corresponds to a down-conversion mixing, the 1-GHz-wide signal is converted to 0 1 GHz. Therefore, the output frequency response of the Gilbert cell needs only be flat up to 1 GHz while the input needs to be able to solve the signal from 3.1 to 4.1 GHz. The design of the multiplier is somewhat identical to that of the VGA. Provided that both circuits are handling a signal of the same frequency properties, only the VGA is explained in more detail. The operation of the Gilbert multiplier is well presented in [37]. The common-mode (CM) voltage at the output of the Gilbert intemultiplier is 2.3 V, which is enough to drive all of the grators. The receiver utilizes a noncoherent energy-collection approach [4], [5], [38]. A detailed analysis of the synchronization process of our receiver is presented in [3]. 4) Link Budget Analysis: In the link budget calculation, we take into account the loss due to the channel. The transmitted satisfies the FCC mask requirements. A signal power level of around 50 mV is desired at the input of the integrators which defines the required receiver chain gain. The desired maximum NF is defined so that the SNR must be more than 10 dB in order to achieve a bit error rate (BER) of 10 [5]. Three possible link budget scenarios for different transmission

1642


TABLE I BUDGET SPECIFICATIONS FOR THREE ALTERNATIVE TRANSMISSION SCHEMES

schemes are presented in Table I, where the gain is the one required in the receiver chain and NF is the maximum allowed noise figure of the receiver chain. The first scheme consists of transmitting 64 pulses per bit and changing the pulse phases at 180 every 16 pulses. The second schemes also utilizes 64 pulses per bit, but the phase of the pulses is changed after each pulse to 180 , in order to further smooth the spectrum due to the overlay scrambling sequence. The third scheme utilizes eight pulses per bit, and the phase is changed for each pulse. We noticed that the first transmission scheme gave us a negative value of the maximum NF at 10 m. This means that we have either to increase the gain in the receiver gain or increase the transmitted power. Since the third scheme has the smoothest spectrum, it provides the most energy when the strongest spectral line is fixed to the maximum allowed transmitted power level.

Fig. 6. UWB testbed.

C. Integrator Architecture and Timing Circuits The applications intended for the UWB application-specific integrated circuit (ASIC) impose low power-consumption requirements so a target of less than 200-mW power consumption for the whole transceiver is intended. We target a maximum power consumption of 8 mW per integrator. The requirement on the output signal swing of the integrators is given by the least significant bit (LSB) of the following analog-to-digital converter (ADC). The higher the LSB is, the bigger the integrator swing must be, and the larger the amplification that is required in the receiver chain. Thus, a lower LSB leads to a smaller power consumption of the transceiver. The power consumption of the integrator bank has the biggest impact on the overall UWB ASIC power consumption. Perhaps the most efficient way to reduce the power consumption of the transceiver is to reduce the power consumption of the transconductor below 3 mW. By reducing the current consumption in the integrator, the output swing and linearity of the integrator will suffer, so a tradeoff must be made between power consumption, output swing, and linearity. The role of the timing circuits as well as the integrator internal structure and receiver analysis are presented in [3]. V. SYSTEM AND CIRCUIT RESULTS The testbed presented in Fig. 6 shows the IR-UWB ASIC mounted on a printed circuit board (PCB) together with the current sources, subminiature A (SMA) connectors, and the RF-balun at the transmitter output and receiver inputs. The measured spectrum of the PJM pulse generator, which is used as the input signal when measuring the implemented receiver spectrum response is presented in Fig. 7. The pulse repetition frequency is 200 MHz. The maximum peak of the spectral

Fig. 7. Measured spectrum at the output of the UWB transmitter (without scrambling sequence overlay).

lines is approximately at 35 dBm. The 200-MHz separation between spectral lines is clearly visible. The measured pulsewidth is 750 ps while the amplitude scale is 2 mV/div. The center frequency of the pulse is 3.2 GHz, and the 10-dB -parameter of pulse bandwidth is 4.7 GHz. The measured the output buffer of the transmitter is presented in Fig. 8. The value is 17.300 dB at 4.27 GHz. The simulated NF of the LNA is presented in Fig. 10. The NF of the LNA has a maximum of 2.7 dB and a minimum of 2.3 dB at the signal bandwidth. The gain of 12 dB is attained with a - and -parameconsumed current level of 10 mA. The -parameter ters of the LNA are presented in Fig. 11. The of the LNA in the signal band is at least 16 dB, which confirms good impedance matching. The -parameter was measured by using the -parameter analysis of the Spectre simulator. The differential input impedance of the LNA is depicted through in Fig. 12. As can be seen, it is very close to 50 whole of the signal band. The 1-dB compression point of the LNA is 3 dBm, with the maximum input signal of approximately 250 mV. This is actually enough to ensure the linearity and dynamic range of the amplifier as long as the waveform


1643

Fig. 10. Simulated NF of the LNA.

Fig. 8. Measured S -parameter at the output of the UWB transmitter.

Fig. 11. Simulated S - and S -parameters of the LNA.

Fig. 9. Measured pulse waveform at the output of the UWB transmitter.

of the signal does not get distorted. According to the transient analysis, the waveform is kept undistorted up to approximately 250 mV. The 1-dB compression point of the VGA is lower than that of the LNA with a maximum input signal of approximately 120 mV. However, this is acceptable, since the maximum input signal level of the LNA is very low. With the LNA gain of 12 dB, the maximum input signal level at the input of the VGA will not exceed the value of a few tens of millivolts. The receiver circuits are implemented on the IR-UWB transceiver chip in two different setups. The first setup includes the VGA, the Gilbert square cell, and the output buffer. The second receiver chain contains the VGA, the Gilbert square cell, a bank integrators, and the FLASH ADC. During the receiver of measurements, the test chain was examined using the HP oscilloscope for waveform visualization and the UWB pulse generator as a transmitter. In Fig. 13, we present the measured spectrum at the output of the first receiver setup which matches the simulated spectrum. The peak-to-peak voltage amplitude of the squared pulses is 83.3 mV, which is enough for the integrator’s

Fig. 12. Simulated differential input impedance of the LNA.

input signal. The performances of the TC and MAX algorithms are compared in terms of TOA estimation. Simulations are performed utilizing the Saleh Valenzuela channel model 3 (CM3) and channel model 4 (CM4), as defined in the IEEE 802.15.4a standard for an indoor office environment [39]. CM3 includes a line-of-sight (LOS) path (corresponding to the shortest TOA) in all channel realizations, but this first path is not always the strongest in the whole impulse response. CM4 is a non-LOS (NLOS) channel model. Perfect synchronization between the

1644


Fig. 13. Spectrum of the tested receiver front-end.

Fig. 15. MAX algorithm T

= 15 ns and E =N = 20 dB. VI. CONCLUSION

Fig. 14. TC algorithm T

= 15 ns, P = 1%, and E =N = 20 dB.

transmitter and receiver clocks is assumed. The simulation pans, ns, which rameters are as follows: ns, and is a random number to the inimplies terval. The number of integrators is considered to be 5, 10, and of 4, 2, 20, respectively, defining the integration windows and 1 ns. For the TC algorithm, the threshold is set using several integration values obtained in the integration window when only noise is present. Both algorithms benefit from an augmented number of integrators, thus reducing the probability of underestimating the delay. On the other hand, the increase in the number of integrators produces a slight increase of the probability of overestimating the delay. From Figs. 14 and 15, it can be noticed that passing from five to ten integrators brings a substantial improvement to the TOA error distribution profile. This improvement is more evident for the TC algorithm than for the MAX algorithm. A further increase in the number of integrators from 10 to 20 brings only a negligible change. Note that, for the TC algorithm, an augmented number of integrators corresponds to a slight increase of the false alarm occurrence.

For UWB systems to become truly ubiquitous in positionestimation-enabled sensor networks, cost, power consumption, and positioning accuracy are critical performance parameters to be tackled. The energy-collection architecture shows great promise for low-cost and low-complexity communications and readily supports TOA estimation. This paper has presented and examined the architecture of an IR-UWB tag transceiver based on a noncoherent energy-collection structure. The receiver chain consists of a UWB LNA with a wideband impedance match to 50 , a Gilbert-cell-based VGA, and a Gilbert cell multiplier used as a squaring circuit. The receiver chain that was implemented has a maximum overall gain of 38 dB, with an overall power consumption of 48 mW and a voltage supply of 3.3 V. An overall NF of 7.3 dB is obtained for the fully differential receiver chain. Measured results of the transmitted spectrum and downconverted received signal are presented. Within the noncoherent architecture, two different TOA algorithms, TC and MAX, have been considered and compared in CM3 and CM4 channel models. With accurate threshold setting and high SNR, the TC algorithm outperforms the MAX algorithm. However, the TC algorithm has relatively high levels of missed detection for low SNR values. The MAX algorithm appears to be more robust to SNR value changes. The MAX algorithm, and its implementation, has been explored in detail. The algorithm has been shown to perform well within the constraints of the low-complexity energy-collection receiver structure. The MAX algorithm implemented can achieve a 5-ns positioning accuracy for 95% of cases. To improve TOA estimation reliability, a current-controlled constant transconductance tuning circuit which preserves the integrator accuracy to within 1% is presented. ACKNOWLEDGMENT The authors would like to thank Prof. T. Rahkonen, Electronics Laboratory, University of Oulu, Oulu, Finland, for his comments, suggestions, and useful discussions, and Stuttgart Design Centre, Thales Electronic Solutions, Stuttgart, Germany, for the help provided during measurements.


REFERENCES [1] “First report and order,” FCC, Washington, DC, FCC 02-48, Apr. 2002. [2] I. Oppermann, L. Stoica, A. Rabachin, Z. Shelby, and J. Haapola, “UWB wireless sensor networks: Uwen—A practical example,” IEEE Commun. Mag., vol. 42, no. 12, pp. S27–S32, Dec. 2004. [3] L. Stoica, A. Rabbachin, H. Repo, S. Tiuraniemi, and I. Oppermann, “An ultra wideband system architecture for tag based wireless sensor networks,” IEEE Trans. Veh. Technol., vol. 54, no. 5, pp. 1632–1645, Sep. 2005. [4] A. Rabbachin, R. Tesi, and I. Oppermann, “Bit error rate analysis for UWB systems with a low complexity, noncoherent energy collection receiver,” in Proc. Int. Telecommun. Symp., Lyon, France, Jun. 27–30, 2004, pp. 223–227. [5] A. Rabbachin and I. Oppermann, “Synchronization analysis for UWB systems with a low-complexity energy collection receiver,” in Proc. Joint Int. Ultra Wideband Syst. Workshop/Ultra Wideband Syst. Technol. Conf., Kyoto, Japan, May 18–21, 2004, pp. 288–292. [6] K. Yu and I. Oppermann, “UWB positioning for wireless embedded networks,” in Proc. IEEE Radio Wireless Conf., Atlanta, GA, Sep. 19–22, 2004, pp. 459–462. [7] ——, “Performance of UWB position estimation based on TOA measurements,” in Proc. Joint Int. Ultra Wideband Syst. Workshop/Ultra Wideband Syst. Technol. Conf., Kyoto, Japan, May 18–21, 2004, pp. 400–404. [8] P. Cheong, A. Rabbachin, M. Jean-Philippe, K. Yu, and I. Oppermann, “Synchronization, TOA and position estimation for low-complexity LDR UWB devices,” in IEEE Int. Ultra-Wideband Conf., Zurich, Switzerland, 5–8, 2005, pp. 480–484. [9] D. Johns and K. Martin, Analogue Integrated Circuit Design. New York: Wiley, 1997. [10] E. A. Homier and R. A. Scholtz, “Rapid acquisition of ultra-wideband signals in the dense multipath channel,” in Proc. IEEE Ultra Wideband Syst. Technol. Conf., Baltimore, MD, May 21–23, 2002, pp. 245–249. [11] R. Fleming, C. Kushner, G. Roberts, and U. Nandiwada, “Rapid acquisition for ultra-wideband localizers,” in Proc. IEEE Ultra Wideband Syst. Technol. Conf., Baltimore, MD, May 21–23, 2002, pp. 105–109. [12] I. Maravic, M. Vetterli, and K. Ramchandran, “Channel estimation and synchronization with sub-Nyquist sampling and application to ultrawideband systems,” in Proc. IEEE Int. Circuits Syst. Symp., Vancouver, BC, Canada, May 23–26, 2004, pp. 381–384. [13] J. Iinatti and M. Latva-Aho, “A modified CLPDI for code acquisition in multipath channel,” in Proc. IEEE 8th Personal Indoor Mobile Radio Commun. Symp., San Diego, CA, Sept. 30–Oct. 3 2001, vol. 2, pp. F6–F10. [14] S. Soderi and J. Iinatti, “CLPDI algorithm in UWB synchronization,” in Proc. IEEE 1st Int. UWB Syst. Workshop, Oulu, Finland, 2–5, 2003. [15] R. Fontana, E. Richley, and J. Barney, “Commercialization of an ultra wideband precision asset location system,” in Proc. IEEE Ultra Wideband Syst. Technol. Conf., Reston, VA, Nov. 16–19, 2003, pp. 369–373. [16] R. J. Fontana, “Recent system applications of short-pulse ultra-wideband (UWB) technology,” IEEE Trans. Microw. Theory Tech., vol. 52, no. 9, pp. 2087–2104, Sep. 2004. [17] C.-H. Yang, K.-H. Chen, and T.-D. Chiueh, “A 1.2 v 6.7 mW impulseradio UWB baseband transceiver,” in Proc. IEEE Int. Symp. Circuits, San Francisco, CA, Feb. 6–10, 2004, pp. 442–443. [18] T. H. Lee, The Design of CMOS Radio-Frequency Integrated Circuits. Cambridge, U.K.: Cambridge Univ. Press, 2004. [19] Y. Wang, J. S. Duster, and K. T. Kornegay, “Design of an ultra-wideband low noise amplifier in 0.13 m CMOS,” in Proc. IEEE Circuits Syst. Conf., Kobe, Japan, May 23–26, 2005, pp. 5067–5070. [20] H.-J. Lee, D. S. Ha, and S. S. Choi, “A systematic approach to CMOS low noise amplifier design for ultrawideband applications,” in Proc. IEEE Circuits Syst. Conf., Kobe, Japan, May 23–26, 2005, pp. 3962–3965. [21] H. C. Schau and A. Z. Robinson, “Passive source location employing intersecting spherical surfaces from tme-of-arrival differences,” IEEE Trans. Acoust., Speech, Signal Process., vol. ASSP-35, no. 8, pp. 1223–1225, Aug. 1987. [22] K. W. Cheung, H. C. So, W. K. Ma, and Y. T. Chan, “Least squares algorithms for time-of-arrival-based mobile location,” IEEE Trans. Signal Process., vol. 52, no. 4, pp. 1121–1128, Apr. 2004. [23] Y. T. Chan and K. C. Ho, “A simple and efficient estimator for hyperbolic location,” IEEE Trans. Signal Process., vol. 42, no. 7, pp. 1905–1915, Jul. 1994.

1645

[24] Y. Huang, J. Benesty, G. W. Elko, and R. M. Mersereau, “Real-time passive source localization: A practical linear-correction least-squares approach,” IEEE Trans. Speech Audio Process., vol. 9, no. 8, pp. 943–956, Nov. 2001. [25] J. Y. Lee and R. A. Scholtz, “Ranging in a dense multipath environment using an UWB radio link,” IEEE J. Sel. Areas Commun. (Special Issue), vol. 20, no. 12, pp. 1677–1683, Dec. 2002. [26] A. Rabbachin, L. Stoica, S. Tiuraniemi, and I. Oppermann, “A low cost, low power UWB based sensor network,” in Proc. Int. Wireless Ad Hoc Networks Workshop, Oulu, Finland, May 31–Jun. 3 2004, pp. 84–88. [27] R. Hoctor and H. Tomlinson, “Delay-hopped transmitted-reference RF communications,” in Proc. IEEE Ultra Wideband Syst. Technol. Conf., Baltimore, MD, May 21–23, 2002, pp. 265–269. [28] J. D. Choi and W. E. Stark, “Performance of ultra-wideband communications with suboptimal receivers in multipath channels,” IEEE J. Sel. Areas Commun. (Special Issue), vol. 20, no. 12, pp. 1754–1766, Dec. 2002. [29] T. Zasowski, F. Althaus, and A. Wittneben, “An energy efficient transmitted-reference scheme for ultra wideband communications,” in Proc. Joint Int. Ultra Wideband Syst. Workshop/Ultra Wideband Syst. Technol. Conf., Kyoto, Japan, May 18–21, 2004, pp. 146–150. [30] J. Foerster and Q. Li, “UWB channel modeling contribution from Intel/” IEEE P.802.15-02/279-SG3a, 2002. [31] S. Tiuraniemi, L. V. Stoica, A. Rabbachin, and I. Oppermann, “Front-end receiver for low power, low complexity noncoherent UWB communications systems,” in Proc. IEEE Int. Ultra-Wideband Conf., Zurich, Switzerland, 5–8, 2005, pp. 339–343. [32] A. Ismail and A. A. Abidi, “An 3–10-GHz low-noise amplifier with wideband LC-ladder matching network,” IEEE J. Solid-State Circuits, vol. 39, no. 12, pp. 2269–2277, Dec. 2004. [33] S. W. Chung, S. Y. Lee, and K. H. Park, “Wideband impedance matching of integrated antennas and CMOS low noise amplifiers for a multiband UWB receiver,” in Proc. IEEE Radio Freq. Integr. Circuits Symp., Atlanta, GA, 19–22, 2004, pp. 131–134. [34] V. Vintola, M. Matilainen, S. Kalajo, and E. Jarvinen, “Variable-gain power amplifier for mobile WCDMA applications,” IEEE Trans. Microw. Theory Tech., vol. 49, no. 12, pp. 2464–2471, Dec. 2001. [35] K. W. Kobayashi, K. T. Ip, A. K. Oki, D. K. Umemoto, S. Claxton, M. Pope, and J. Wiltz, “Gaas HBT 0.75-5-GHz multifunctional microwave-analog variable gain amplifier,” IEEE J. Solid-State Circuits, vol. 29, no. 10, pp. 1257–1261, Oct. 1994. [36] R. G. Meyer and W. D. Mack, “A dc to 1-GHz differential monolithic variable-gain amplifier,” IEEE J. Solid-State Circuits, vol. 26, no. 11, pp. 1673–1680, Nov. 1991. [37] B. Gilbert, “A precise four quadrant multiplier with sub nanosecond response,” IEEE J. Solid-State Circuits, vol. 3, pp. 365–373, Dec. 1968. [38] L. Stoica, H. Repo, S. Tiuraniemi, and I. Oppermann, “An ultra wideband low complexity circuit transceiver architecture for sensor networks,” in Proc. IEEE 2005 Circuits Syst. Conf., Kobe, Japan, May 23–26, 2005, vol. 1, pp. 364–367. [39] “IEEE-802.15.4a channel model—Final report,” IEEE-802.15.4a, 2004 [Online]. Available: http://www.ieee802.org/15/pub/TG4a.html

Lucian Stoica (S’01) was born in Roman, Romania, on December 8, 1975. He received the M.S. degree in electrical engineering from the Technical University of Iasi, Iasi, Romania, in 2000, and is currently working toward the Ph.D. degree in electrical engineering at the University of Oulu, Oulu, Finland. From 2000 to 2003, he was a Teaching Assistant with the Telecommunications Department, Technical University of Iasi, where he was involved in the development of digital design and FPGA prototyping. In 2003, he joined the Centre for Wireless Communications, University of Oulu. During 2005, he was a Visiting Researcher with Thales Electronic Solutions, Stuttgart, Germany. His current research focuses on low-complexity SiGe BiCMOS circuit transceivers design for wireless communications, particularly on UWB impulse-radio systems.

1646


Alberto Rabbachin (S’04) received the M.S. degree from the University of Bologna, Bologna, Italy, in 2001, and is currently working toward the Ph.D. degree at the Universityof Oulu, Oulu, Finland. In 2001, during his undergraduate studies, he visited the Centre for Wireless Communications, University of Oulu, Oulu, Finland. In 2002, he joined Agilent Technologies for an internship and, since 2003, he has been with the Centre for Wireless Communications, University of Oulu. During 2005, he was a Visiting Researcher with CEA-LETI, Grenoble, France. His research interests include UWB systems with emphasis on receiver structures, synchronization, and ranging techniques.

Ian Oppermann M’91–SM’02) was born in Maryborough, Australia, in 1969. He received the B.Sc., B.E., and Ph.D. degrees from the University of Sydney, Sydney, Australia, in 1990, 1992, and 1997, respectively. His doctoral research concerned physical-layer aspects of novel spread-spectrum/CDMA systems. In 1996, he founded SP Communications, a company that developed network planning tools for thirdgeneration mobile systems and IP cores for wireless local area network (WLAN) chipsets. In 2001, he became a Docent (Adjunct Professor) with the University of Oulu, Oulu, Finland. In 2002, he joined the Centre for Wireless Communications (CWC), initially as an Assistant Director, and then becoming Director in 2003. Since the beginning of 2005, he has been the Acting Director for Short Range Communications Research, CWC. His main research interests are spread-spectrum systems and UWB. He coedited several books. He has authored or coauthored over 80 publications in international journals and conferences. He holds several patents for wireless communications.