The present disclosure relates to a short-range wireless receiver with near-threshold digital baseband.
An exponential growth in miniaturized smart sensors is imminent in the near future. The rapid growth in technology is bringing the vision of Internet of Things (IoTs) closer to reality at a much faster pace than previously anticipated. This value does not come by connecting every object to the internet but by their intelligent interaction and collaboration. This will open new dimensions of collecting data and extracting information at a scale not possible before. This technology will enable smart cities with improved waste/water management, transportation and lighting, connected cars with smart homes and will revolutionize retail, manufacturing, shopping and healthcare.
The sensor density around a person is expected to increase from a few hundreds to thousands, which will correspond to roughly a trillion networked sensors on the planet. The microsystems encompassing these sensors will have to have high-energy efficiency for computation, communication and sensing operations. This is mainly because many of these microsystems are expected to operate at the edge of the cloud with a battery lifetime of 10+ years, or batteryless operation from harvested energy. This poses new design challenges and opportunities for circuit designers and especially for wireless communication Integrated Circuits (ICs) as they consume a significant amount of power when active in a miniaturized microsystem.
Some efforts are put in place to define an open platform that enables the Internet of Things such as 6LoWPAN-based networks built on the IEEE 802.15.4 standard. Recently, IEEE 802.15.4 compliant radio frequency (RF) front-ends have been reported with exceptional sensitivity (wireless range>100 m) and energy efficiency of 7.2 nJ/bit, 6.8 nJ/bit and 7.4 nJ/bit.
However, there are many IoT applications that only require short-range communication (<10 m), such as wireless proximity sensors for smart meters and parking spaces, home automation within a room, and some wearables for fitness and health monitoring. In these cases, different design tradeoffs can be made in order to improve energy efficiency, as compared to devices that prioritize high performance or ICs designed for worst-case applications. In particular, it is well known that the sensitivity of a receiver directly trades off with its power consumption. Dialing sensitivity back to around −50 dBm could lower the power of the radio significantly and meet the needs of many energy-constrained applications. However, doing so is not trivial, and requires redesigning with focus on ultra-low power from RF front-end through the digital baseband processor.
This disclosure presents a fully integrated 2.4 GHz receiver comprising an RF front-end, analog-to-digital converter (ADCs), and digital baseband processor (DBB) that exploit the relationship of sensitivity and power consumption by adapting the sampling and processing rates of signals in the radio baseband processor. Although not meeting the sensitivity required by the IEEE 802.15.4 standard, this receiver provides a short-range O-QPSK DSSS link that is fully compatible with IEEE 802.15.4 packets. While reference is made throughout this disclosure to a particular standard, it is readily understood that the concepts described herein are applicable more generally to short-range wireless receivers.
This section provides background information related to the present disclosure which is not necessarily prior art.
This section provides a general summary of the disclosure, and is not a comprehensive disclosure of its full scope or all of its features.
A method is provided for operating a short-range receiver with a digital baseband processor. The method includes: receiving a network data packet over a data link of a wireless network; sampling a portion of the network data packet (e.g., header) at a given sampling rate (e.g., Nyquist rate); determining a metric indicative of quality of the data link over which the network data packet was received; lowering the given sampling rate of the network data packet to a reduced sampling rate, where the reduced sampling rate is set inversely to quality of the data link and with a value lower than the given sampling rate; and processing remainder of the network data packet at the reduced sampling rate.
In one embodiment, the metric is compared to a threshold, the given sampling rate is reduced to the reduced sampling rate and the remainder of the network data packet is processed when the metric is above a threshold; whereas, the remainder of the data unit is processed at the given sampling rate when the metric is below the threshold.
Lowering the given sampling rate may be implemented by disabling one of the digital baseband process, an analog-to-digital converter or another active circuit in the receiver for a period of time within the sampling interval.
In some embodiments, lowering the given sampling rate includes ranking samples from the portion of the network data packet according to energy level; selecting a subset of the samples having highest energy level, where the number of selected samples correlates to the reduced sampling rate; and sampling the remainder of the network data packet in accordance with the selected subset of samples.
In some embodiments, the metric indicative of quality of the data link may be selected from a group comprised of a signal-to-noise ratio, a received signal strength index and a link quality indicator.
In one aspect of this disclosure, the method for operating a short-range receiver can be further defined as follows. The method includes: receiving, by the receiver, a protocol data unit over a channel in a wireless network; sampling, by the receiver, a portion of the data unit at a given sampling rate; determining, by the receiver, a metric indicative of quality the channel over which the data unit was received; comparing, by the receiver, the metric to a threshold; processing the remainder of the data unit at a reduced sampling rate when the metric is above the threshold; and processing the remainder of the data unit at the given sampling rate when the metric is below the threshold. Processing of the data unit further includes ranking samples from the portion of the data unit according to energy level; selecting a subset of the samples having highest energy level, where the number of selected samples correlates to the reduced sampling rate; and sampling the remainder of the network data packet in accordance with the selected subset of samples.
In another aspect of this disclosure, a short-range receiver is provided. The receiver includes: an RF front-end circuit, and analog-to-digital converter and a digital baseband processor. The RF front-end circuit is configured to receive an RF analog signal from an antenna and operates to shift the RF analog signal to an intermediate signal having a different frequency. The analog-to-digital converter is configured to receive the intermediate signal from the RF front-end circuit and convert the intermediate signal to a digital signal. The digital baseband processor is configured to receive the digital signal from the analog-to-digital converter and processes the digital signal at a prescribed sampling rate. More specifically, the baseband processor includes a link quality module that determines a metric indicative of quality of data link over which the RF analog signal was received and set the prescribed sampling rate to a value that is inversely related to the quality of the data link; and a decoder that samples the digital signal at the prescribed sampling rate and outputs a sequence of data bits derived from the digital signal.
Further areas of applicability will become apparent from the description provided herein. The description and specific examples in this summary are intended for purposes of illustration only and are not intended to limit the scope of the present disclosure.
The drawings described herein are for illustrative purposes only of selected embodiments and not all possible implementations, and are not intended to limit the scope of the present disclosure.
Corresponding reference numerals indicate corresponding parts throughout the several views of the drawings.
Example embodiments will now be described more fully with reference to the accompanying drawings.
To understand system design tradeoffs for RF receivers consider a generic radio front-end as shown in
High performance and high sensitivity receivers (sensitivity<−90 dBm) require a low receiver Noise Figure (NF). This is achieved by employing a Low Noise Amplifier (LNA) at the front of the signal-processing path, followed by additional RF gain stages and an active mixer. The LNA noise figure typically dominates the receiver overall noise figure and trades directly with power. The noise factor (F) of an LNA is related to its power approximately by,
where PLNA is the power consumed by a low-noise amplifier and α is the proportionality constant that depends on the given technology and the circuit topology.
The overall linearity of the RF front-end is dictated by the baseband gain stages. To better understand the linearity vs. power tradeoff, a three-point method is adopted to estimate the linearity of a short-channel NFET in 65 nm CMOS. For a zero-IF receiver architecture second-order linearity is more important and the same method is used to estimate IIP2.
where IIP2, IIP3 are the second and third order input intercept points respectively, g is the incremental device gain evaluated at three input voltages 0, V and −V, and Rs is the source resistance. The incremental gain of a short-channel MOSFET is given by,
where Vod is the over-drive voltage and ρ takes velocity saturation into account. The Vod depends only on the current density therefore both IIP3 and IIP2 are plotted vs. current density, to estimate power in 65 nm CMOS as shown in
Apart from the sensitivity vs. power tradeoff, applicant proposes an adaptive sampling method for operating the receiver as seen in
From the sampled data, a quality metric is determined at 43, where the quality metric is indicative of quality of the data link over which the network data packet was received. In the example embodiment, signal-to-noise ratio is used as the quality metric. Other types of quality metrics are contemplated by this disclosure including a received signal strength index and a link quality indicator.
To save power, the receiver can be operated at a lower sampling rate while maintaining link performance. In a simplified example, the quality metric is compared at 45 to a threshold. When the quality metric exceeds the threshold, the sampling rate can be lowered at 46, thereby reducing power consumption. For example, the sampling rate can be set at 50% or 25% of the standard rate. In other words, the sampling rate is set inversely to the quality of the data link. The remainder of the data packet is then processed at the reduced sampling rate as indicated at 27. The reduced sampling rate can be implemented in different ways as further described below.
When the quality metric is less than or equal to the threshold, the sampling rate remains the same and the remainder of the data packet is processed at the standard sampling rate. Upon receiving another data packet, the process is repeated as indicated at 48. In this way, adaptive sampling is performed on a packet-by-packet basis. It is to be understood that only the relevant steps of the methodology are discussed in relation to
In a variant of this method, processing of additional data packets continues at the reduced sampling rate until the quality of the data link changes. That is, the quality of the data link is monitored by the receiver. Upon detecting a change in the quality of the data link, the sampling rate is set in the manner set forth above.
Rather than arbitrarily selecting samples at the reduced sampling rate, samples are selected in a particular manner. In the example embodiment, once the averaged channel pulse template is learned, samples are ranked with respect to energy level and this information is later used for adapting the average sampling rate. More specifically, a subset of samples (having the highest energy level) is selected and the remainder of the data packet is sampled in accordance with the subset of samples. With continued reference to
In one embodiment, the digital baseband processor will only process the selected samples and cease operating for the other samples. In the case of the 25% sampling rate, the digital baseband processor would process the third pulse but not the other three pulses. Likewise, other components of the receiver can be disabled for a period of time within a sampling interval according the reduced sampling rate. For example, an analog-to-digital converter can be disabled except to process the third pulse in the case of a 25% sampling rate. One or more other active circuits, such as amplifiers, filter, oscillators, etc., can also be disabled for a period of time within the sampling interval on a per-sample basis according to the reduced sampling rate. It is also understood one or more of these steps can be taken in combination to reduce power consumption.
To evaluate the link performance using the proposed adaptive sampling, a MATLAB simulation model was developed.
In the example embodiment, an RF signal at 2.45 GHz is directly fed into an active Gilbert cell based mixer 73 and quadrature down-converted to baseband. Channel selection is performed by a third-order Butterworth gm-C active low-pass filter 74 with a corner frequency of 1.5 MHz. The filtered baseband signal is then amplified by three Programmable Gain Amplifiers (PGAs) 75 and followed by a buffer 76 which drives the input of the flash ADC. The 5-bit flash ADC 78 samples the incoming I & Q baseband signals at 4 MHz (2× the required Nyquist rate). Open loop digital DC-offset calibration is distributed across the front-end by using current DACs (Digital to Analog Converters) in the active filter and PGAs. The I & Q baseband chips are then processed by the digital baseband processor 79.
From simulated BER curves, the 5-bit resolution of the flash ADC is determined to have negligible impact on the link performance. The comparator offset in a flash ADC is reduced to be less than LSB/4 by sizing the transistors of the input stage. A MATLAB model has been developed to evaluate the ENOB (Effective Number Of Bits) of the flash ADC for the comparator offset measured from Monte Carlo simulations.
The proposed receiver is intended for short-range wireless communication so to extend the communication range the wireless nodes are expected to operate in a mesh network. The Friis equation (6) below can be used to calculate the maximum line-of-sight communication range between two sensor nodes. The measured receiver sensitivity from RF-to-bits at 10−3 BER is −52.5 dBm. Using the ISM band center frequency of 2.45 GHz and assuming a +7 dBm EIRP transmitter, the communication range corresponding to −52.5 dBm RX sensitivity is found to be 9.2 m.
To obtain the desired performance, the noise figure (NF), linearity and gain requirements of the receiver are calculated and summarized in Table 1 below.
The Packet Error Rate (PER) is related to the Symbol Error Rate (SER) if acquisition effects are ignored by the relation.
For IEEE 802.15.4, the number of bits in a packet is N=160 bits with 48 bits of overhead, this corresponds to 52 symbols in a packet and therefore 1% PER corresponds to 0.019% SER. A single symbol error would result in, on average, k/2 bit errors, where k=4 is the number of bits in a symbol. This corresponds to 0.0095% BER. The BER of O-QPSK modulation with half-sine pulse shaping is given by the Q-function (8) below.
Therefore for 0.0095% BER the required (Eb/No)min would be 8.8 dB.
Direct Sequence Spread Spectrum (DSSS) adds a Coding Gain (CG) and Processing Gain (PG). The coding CG is related to the degree of orthogonality for the code set, which for the DSSS code is calculated from the mean Hamming distance of the code set,
where n is the length of the code. The CG is approximately 2 dB which reduces the required (Eb/No)min to 6.8 dB. The Processing Gain is calculated by the ratio of the chip rate to the data rate.
The chip rate is 2 Mcps and data rate is 250 Kbps that corresponds to PG of about 9 dB. The PG doesn't reduce the energy per bit required in contrast to CG but it's rather a measure of how much more energy is used to detect a bit as compared to the energy per chip (ES). Hence the minimum (ES/NO)min required to achieve 1% PER considering CG and PG can be calculated which is −2.2 dB for the Nyquist rate sampling and about 0 dB for 50% sampling computed from MATLAB simulations as shown in
For other receiver performance parameters, the ADC reference voltage is 300 mV, reference impedance is 50 Ohms, insertion loss for the RF band select filter is assumed to be 2 dB and the link margin is 10 dB. The NF of the receiver front-end is calculated by where BW is assumed to be 1.5 MHz.
where Rss is the target receiver sensitivity.
The IEEE 802.15.4 standard doesn't specify the linearity requirements of the receiver front-end. Hence the linearity requirements can be derived from the interferer profile. IIP3, IIP2 and SFDR are calculated as follows,
where Pint is the power of the interferer and Psig is the power of the desired signal.
F is the receiver noise factor.
The maximum and minimum gain required from the front-end assuming 5-bit ADC (NADC) and a Back Off (BO) margin of 10 dB is calculated as follows,
Gmax=REFADC−6NADC+SNRmin−RSS+Margin (15)
Gmin=REFADC−Rmax−BO (16)
where Rmax is the maximum received power that is −20 dBm and REFADC is the ADC reference voltage power relative to 50 Ohm. The theoretical link budget along with measured performance is shown in Table 1.
In an example embodiment, the link quality module 96 implements the adaptive sampling technique set forth above. That is, the link quality module 96 determines a metric indicative of quality of data link over which the RF analog signal was received and sets the prescribed sampling rate to a value that is inversely related to the quality of the data link. In the example embodiment, the 4 MHz sampling rate corresponds to four samples per I & Q symbol. From the computed channel pulse response, the link quality module 96 ranks the four samples with respect to energy. This is conceptually illustrated in
In one embodiment, digital clock gating is used in the digital baseband at the module level to save computational power. The digital clock is enabled for a module if it needs to process the current time sample from the ADC. The time samples are ranked in energy from the learned channel pulse template. The clock is enabled for a module only for the time samples corresponding to the highest energy samples consistent with the selected average sampling rate. In this way by reducing the average frequency of the digital processing unit, the total average power is reduced
A decoding module 97 uses a matched-filter to perform hard decision decoding (HDD). HDD is used for lower computational complexity as compared to soft-decision decoding (SDD), with roughly a 2 dB penalty in link performance. Finally, the de-spreader module 98 despreads the received chips and outputs the raw binary data transmitted. For BER testing, the digital baseband enters into a state where it demodulates the data infinitely. Module level clock gating is used in the digital baseband to save power. As used herein, the term “module” may refer to, be part of, or include: an Application Specific Integrated Circuit (ASIC); a digital, analog, or mixed analog/digital discrete circuit; a digital, analog, or mixed analog/digital integrated circuit; a combinational logic circuit; a field programmable gate array (FPGA); a processor circuit (shared, dedicated, or group) that executes code; a memory circuit (shared, dedicated, or group) that stores code executed by the processor circuit; other suitable hardware components that provide the described functionality; or a combination of some or all of the above, such as in a system-on-chip.
The LO buffer is a resistive loaded differential amplifier cascaded with a source follower for a DC level shift. To save power, the receiver doesn't use an LNA and instead relies on an active mixer to provide RF gain. Since the baseband modulated signal has significant low-frequency content, the devices in the active mixer are sized to reduce the flicker-noise corner frequency to <100 KHz.
The IEEE 802.15.4 PHY requires 0 dB rejection at the adjacent channel (±5 MHz) and 30 dB rejection at the alternate channel (±10 MHz). Assuming 10 dB margins, 40 dB rejection at the alternate channel can be achieved through the third-order Butterworth-type filter with corner frequency of 1.5 MHz. The filter will provide 50 dB rejection at 10 MHz apart from the wanted signal and thus can be used as the channel selection anti-aliasing filter. To adjust the corner frequency of the filter over process corners, the capacitors are made tunable by a 3-bit binary control word to vary capacitance by ±20%. The differential output of the mixer is converted into single-ended by the input stage of the gm-C filter.
In the example embodiment, the entire baseband is implemented single-ended to save power. The baseband gain is distributed between the active filter and the PGAs. Programmable gain (PG) is implemented by switchable fixed gain-stages.
The gain stage is implemented as a modified first order gm-C stage. A transmission gate is used which when enabled allows the input signal to bypass the gain stage which is being disabled by a footer. Each PG stage provides a gain of about 8 dB for a total PGAs' gain of roughly 24 dB. For distributed offset calibration, the current DACs are designed to reduce the DC-offset to within LSB/2 of the flash ADC. The output of the PG stages is fed into a buffer that drives the input capacitance of the flash ADC. S/H circuit is avoided at the input of the flash ADC considering that 1 MHz baseband signal isn't fast enough relative to the comparator speed in 65 nm CMOS to cause aperture errors. The LSB size is 9.4 mV for a reference voltage of 300 mV, generated off-chip. To reduce power, no pre-amplifier is used in the comparator that makes the flash converter susceptible to comparator kickback. To reduce comparator kick-back and the power consumption of the reference ladder, decoupling capacitors of 2 pF are added to the reference ladder as shown in
The example receiver described above is tested with IEEE 802.15.4 RF compatible packets. An off-chip LO power of −6 dBm and a 90° hybrid coupler is used to generate the quadrature LO signals. An FPGA is used to configure the scan-chain.
For a BER test, the digital baseband enters into a state where it receives the data infinitely. The measured sensitivity of the RX is −52.5 dBm at 10−3 BER. From the measured BER performance, it is observed that if the input SNR is about 3 dB higher at 2× the Nyquist sampling than for the same link performance of 10−3 the DBB can be operated at 25% samples with an energy efficiency of 2.1 nJ/bit.
For a complete solution, the receiver would require an on-chip LO. In order to estimate total power, we draw from previously published LOs. A 2.4 GHz LO generated using a PLL consumed 1.6 mW including the PLL, LC-VCO and VCO buffer and 1.2 mW for an entire QVCO [23]. This corresponds to roughly total power of 3 mW for a complete receiver, including this work. This receiver has 2× better energy efficiency for the radio front-end (3.5 nJ/bit) than previous approaches, while reporting 8.1 nJ/bit energy efficiency for an O-QPSK DSSS coherent receiver with near-threshold digital baseband.
Radio energy efficiency is going to play a key role in extending the battery life of future IoT devices. Different design tradeoffs can be made for these emerging applications as compared to the conventional high performance radios. We have explored the sensitivity vs. power tradeoff and have presented a low-power (2 mW) short-range O-QPSK DSSS receiver with integrated digital baseband. The radio receiver also adapts its average sampling rate for high input SNR while still maintaining the target link performance of 10−3 BER to save power in the digital baseband.
The foregoing description of the embodiments has been provided for purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosure. Individual elements or features of a particular embodiment are generally not limited to that particular embodiment, but, where applicable, are interchangeable and can be used in a selected embodiment, even if not specifically shown or described. The same may also be varied in many ways. Such variations are not to be regarded as a departure from the disclosure, and all such modifications are intended to be included within the scope of the disclosure.
This application claims the benefit of U.S. Provisional Application No. 61/984,316, filed on Apr. 25, 2014. The entire disclosure of the above application is incorporated herein by reference.
This invention was made with government support under CCF0910765 awarded by the National Science Foundation. The Government has certain rights in the invention.
Number | Name | Date | Kind |
---|---|---|---|
7907555 | Sankabathula | Mar 2011 | B1 |
20020080898 | Agazzi | Jun 2002 | A1 |
20040110510 | Jeon et al. | Jun 2004 | A1 |
20050195920 | Gierl | Sep 2005 | A1 |
20060104339 | Langberg | May 2006 | A1 |
20060188040 | Vis | Aug 2006 | A1 |
20060193371 | Maravic | Aug 2006 | A1 |
20060222130 | Dominique | Oct 2006 | A1 |
20090279650 | Qian | Nov 2009 | A1 |
20090311985 | Youssoufian et al. | Dec 2009 | A1 |
20110096875 | Amrutur | Apr 2011 | A1 |
20120257558 | Shin | Oct 2012 | A1 |
20120257897 | Hu | Oct 2012 | A1 |
20130019685 | Sakaguchi et al. | Jan 2013 | A1 |
20130243069 | Guo | Sep 2013 | A1 |
20140241157 | Tan | Aug 2014 | A1 |
20150282068 | Rajagopal | Oct 2015 | A1 |
Number | Date | Country |
---|---|---|
1635331 | Mar 2006 | EP |
2467984 | Jun 2012 | EP |
WO-2010143756 | Dec 2010 | WO |
Entry |
---|
S. Lanzisera et al “Reducing Average Power in Wireless Sensor Networks Through Data Rate Adaptation”, IEEE ICC, 978-1-4244-3435 (2009). |
Number | Date | Country | |
---|---|---|---|
20150311946 A1 | Oct 2015 | US |
Number | Date | Country | |
---|---|---|---|
61984316 | Apr 2014 | US |