BMe Research Grant
Wireless networking devices of embedded systems must satisfy special requirements that cannot be met by further improving the performance of mobile phone type systems. Since one device should operate for years on the same battery and its price should only be a few US dollars, CMOS technology must be used. As there are no more empty frequency bands available for the CMOS devices to operate, the reuse of already occupied frequency bands is inevitable. To avoid interference, the power spectral density of radiated signal must be kept very low. To spread the spectrum of transmitted signal, ultra-wideband systems do not use sinusoidals as carriers but pulses or chaotic signals instead. However, these carriers have no amplitude, frequency or phase in the conventional sense.
I have been conducting my research at the Department of Measurement and
Information Systems as a PhD student. The Department provides all the necessary
professional, technical and moral support for the researchers to build up the
most efficient research environment. For my research, an RF PXI platform donated
by the National Instruments is available, therefore it is possible to generate real
physical waveforms with arbitrary envelope. In this way, additionally to the simulation results I could use measurement data to validate my theoretical
There are two emerging applications where brand new wireless networking devices are required: (i) sensor networks and (ii) embedded systems. Nodes of sensor networks collect data from the area they monitor and after some data pre-processing and merging, the information is transmitted to a central unit. Sensor networks are used everywhere, such as environmental monitoring, high precision agriculture, detection of chemical agents, health care, etc.
An embedded system is a dedicated distributed computer system that actively controls a whole process beyond measuring process parameters. Areas of typical applications of embedded systems include automatic manufacturing lines, smart houses, control of shopping malls, smart offices, etc.
Both sensor networks and embedded systems are large-scale distributed computer systems that involve many nodes. Each node participates in the collection, pre-processing and merging of data as well as controlling the process monitored by the system. The required flexibility, easy installation and reconfiguration can only be delivered by wireless networking devices.
Since available radio channels have already been allocated to conventional radio communication systems, frequency reuse is the only solution to accommodate new radio channels. Frequency reuse, however, cannot be assured by enhancing conventional communication technology, therefore a brand new approach, using the ultra-wideband (UWB) radio is required . Taking the UWB approach, during my research I develop new wireless communication technologies for the sensor network and embedded system applications both at protocol and hardware levels.
Design constraints of wireless networking devices of embedded
systems cannot be met by further improving the performance of currently used
communications systems. To satisfy the new demands, the following problems must be solved by basic research:
1. Development of a general comprehensive theory for the UWB networking devices of embedded systems and waveform communications
2. Derivation of analytical tools and closed-form equations for the noise and interference performance of UWB networking devices
3. Evaluation of the effect of channel conditions on system performance
4. Development of a general comprehensive theory for the modulation schemes used in UWB networking devices
5. Based on the complex envelope approach, elaboration of new detection algorithms, signal processing and detector configurations that can be used in zero-IF receivers built with SoC CMOS technology
Extension of the Fourier analyzer concept to UWB impulse radio. The generalized theory will cover each type of digital communications.
Elaboration of a scalar measure for the amount of a priori information
Installation of a computer controlled measurement test bed based on 2.7 GHz vector signal generator and analyzer units. Development of test software for the PHY layers of networking devices
- General theory of waveform communications
- Measure for a priori information
- Test bed for networking devices of embedded systems
Derivation of a unified mathematical framework that covers each kind of UWB modulations
Based on the theory developed, elaboration of analytic methods for derivation of UWB detection algorithms
Using the analytic tools developed, elaboration of new detection algorithms for impulse and chaos-based UWB systems
- Mathematical framework and analytical tools for the development of UWB modulation schemes and detection algorithms
- New detection algorithms
Derivation of analytical tools for performing the noise performance analysis of UWB networking devices
Development of closed-form equations for AWGN noise performance of UWB networking devices
Using the tools developed, optimization of noise performance of UWB networking devices under different channel conditions
- Analytical tools for noise performance analysis
- Closed-form equations for the noise performance evaluation of UWB networking devices
- Determination of effect of channel conditions on UWB system performance
Evaluation and comparison of interference performance of different UWB networking devices
Based on the theory of complex envelopes, development of new receiver configurations and detector circuits for UWB networking devices suitable for SoC CMOS technology
- Analytical tools for the interference analysis
- New detection and signal processing circuits for UWB networking devices
In traditional communication systems where the carrier fills up entirely the bit duration the link budget is based on the SNR (Signal-to-Noise-Ratio) . In the case of low-data rate UWB impulse radio this approach cannot be used because the duty cycle is typically very low and the PAPR (Peak-to-Average Power Ratio) is high. I suggest using a link budget calculation relying on the bit energy. The coverage of UWB impulse radio is limited by the FCC  Regulations that determine the bit energy. There is a further limitation on the bit energy, namely, the low voltage level of UWB applications implemented with CMOS technology. I derived the link budget calculation according to the new bit energy-based approach and showed that the coverage of impulse radio applications used indoor was below 2 m in the case of NLOS (Non-Line of Sight) scenario. This value agrees with the measurement results in two research projects, the PULSERS and the EUWB financed by the EU. Owing to the traditional approach taken, they could not explain the low coverage obtained. Since low bit-energy is a consequence of the extremely narrow impulses (some nanosecs), I showed how the bit energy could be improved without violating the FCC Regulations. IEEE802.15.4a Standard  allows using chaotic waveforms as carriers in UWB systems. I propose using FM modulated chaotic waveforms as carriers to increase the bit energy which have longer duration than impulses. While ultra-wideband property for the impulse radio is attributed to the very short pulse duration, chaotic signal is inherently wide band, i.e. the duration of the chaotic waveform has no influence on the bandwidth if certain conditions are met. Another important property of the chaotic (and impulse) waveforms is that they can be generated by simple circuitry. Using a chaotic carrier in an UWB system can increase bit energy by 16 dB and 20 dB compared to an UWB impulse radio system if the duration of the chaotic UWB waveform is increased to 300 ns and 800 ns, respectively. This increase in bit energy improves coverage considerably.
The FCC Regulations contain no restrictions on the carrier or the modulation scheme used in the UWB systems; they only regulate the constraints under which the peak and power limits of the transmitted waveforms have to be satisfied. The FCC Regulations also describe the method of power measurements. It is important to emphasize that the power limits provided there should always be measured at the output of the required FCC filter, so the limits have no direct relevance to the transmitted waveform. The waveforms I applied in my research had Gaussian envelope. The Regulations specify a minimal bandwidth of 500 MHz for the UWB waveforms . This bandwidth is much wider than that of the FCC filter. Therefore, such a narrow UWB impulse can be considered a Nascent-delta function when exciting the FCC filter input by such a waveform. As a consequence, the impulse response of the filter can be measured at the FCC filter output. I had validated this idea by simulations and then constructed the baseband model of the FCC measurements for both peak and average limits. The low-rate UWB impulse radio system is peak power limited. Starting from the FCC peak power limitation I derived the value of the highest peak amplitude that determines the coverage in the case of impulse-based communication. Since the coverage provided by a single impulse is too low, I propose the transmission of multiple pulses per bit in burst while keeping the FCC Regulations. However, we have to consider the impulse response at the output of the FCC filter and the impulses in the burst have to satisfy the FCC limit even if in-phase superposition between the pulses in the burst occurs. I described the relationships between the parameters of the UWB burst and the voltage levels at the FCC filter output. I clarified the effect of changing the time delay between the impulses in the burst and examined the effect resulting from detuning the filter center frequency.
The transmitted reference UWB impulse radio with autocorrelation receiver, the pulse position modulated and on-off keying modulated UWB impulse radio with energy detector offer simple and robust solutions for low cost, low data rate, low power consumption wireless applications . Unfortunately, relatively low sensitivity of built UWB receivers limits the attainable radio coverage. An improvement in noise performance is a must. The duty cycle of the UWB impulse radio is extremely low. I propose to exploit this feature to increase noise performance. Noise performance is strongly dependent on the product of receiver bandwidth 2B and energy capture time \tau. The lower the product, the better the noise performance and the better the receiver sensitivity will be. I showed that optimal matching of the parameters of the autocorrelation receiver and energy detector-based receivers to the received waveform significantly increases the noise performance in the case of a Gaussian waveform.
expression, where E_b is the bit energy,
N0/2 denotes the power spectral density (psd) of channel noise, 2B is the
noise bandwidth determined by the channel selection filter and \tau is the
energy caption time, valid for the calculation of the bit error ratio in the case of
the autocorrelation receiver and energy detector type receivers. The analytical
expression is only valid for integer 2B\tau values. In the first step I
searched for an optimum using the above expression, then continued
searching the optimum for non integer 2B\tau values using a validated Matlab simulator built by myself. As a result, I managed to achieve 7.7 dB improvement in noise performance. I proved validity of the above expression for
UWB transmitted reference, pulse position modulated and on-off keying modulated
system as well, provided that each systems had identical bit energy.
Considering CMOS implementability, transmitted reference system is the
best choice as one requiring the lowest voltage level for the same
amount of bit energy.
A comprehensive theory of UWB communications cannot be found in the literature. As a result, the UWB modulation schemes and detector configurations available today have been developed using heuristic approaches, so the detection algorithms cannot be optimized and matched to the channel condition. Chaotic signals could be used as UWB carriers but the lack of a comprehensive theory of UWB systems prevents the comparison with different carriers.
An expected result from my research, cost-effective implementation of CMOS UWB networking devices will be elaborated. The applicability and performance of chaotic carriers in UWB applications will be evaluated and compared to that of the UWB impulse carriers.
Expected results are as follows:
1) A comprehensive theory covering both the impulse and chaotic carriers will be developed;
2) New modulation schemes and detection algorithms for UWB communications will be derived;
3) New UWB receiver configurations will be elaborated;
4) Chaotic signal generator suitable for the generation of UWB carrier will be developed;
5) Acknowledgment protocols for LR-WPAN applications will be elaborated.
List of the applicant's publications for download: http://home.mit.bme.hu/~krebesz/publications.html
List of publications and citations in PDF format: http://mit.bme.hu/~krebesz/kutatoi_bme/publication.pdf
Selected publications: http://mit.bme.hu/~krebesz/kutatoi_bme/selected_publications.pdf
List of professional results: http://mit.bme.hu/~krebesz/kutatoi_bme/szakmai_eredmenyek02.pdf
Brief professional CV: http://mit.bme.hu/~krebesz/kutatoi_bme/krebesz10cv_szakmai02.pdf
Letter of support from the supervisor: http://mit.bme.hu/~krebesz/kutatoi_bme/ajanlas.pdf
 Federal Communications Commission, Part 15 of the Commission Rs Rules Regarding Ultra-Wideband Transmission Systems; Subpart F, FCC–USA, Online: <http://sujan.hallikainen.org/FCC/FccRules/2009/15/>
 IEEE Standard 802.15.4a-2007, IEEE Computer Society, LAN/MAN Standards Committee, Work Group 15, Task Group 4a, 2007
 P. P. Mercier, D. C. Daly, M. Bhardwaj, D. D. Wentzloff, F. S. Lee and A. P. Chandrakasan: “Ultra-low-power UWB for sensor network applications," in ISCAS’08, Seattle, Washington, USA,May 18–21 2008, pp. 2562–2565
 S. Haykin: Communication Systems, 3rd ed. John Wiley & Sons, 1994
 K. Witrisal et. al.: “Noncoherent Ultra-Wideband Systems:
An Overview of Recent Research Activities,” IEEE Signal
Processing Magazine, vol. 26, no. 4, pp. 48–66, July 2009