Article

Neural signal compression using a minimum Euclidean or Manhattan distance cluster-based deterministic compressed sensing matrix

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Multichannel wireless neural signal recording systems are a prominent topic in biomedical research, but because of several limitations, such as power consumption, the device size, and enormous quantities of data, it is necessary to compress the recorded data. Compressed sensing theory can be employed to compress neural signals. However, a neural signal is usually not sparse in the time domain and contains a large number of similar non-zero points. In this article, we propose a new method for compressing not only a sparse signal but also a non-sparse signal that has identical points. First, several concepts about the identical items of the signal are introduced; thus, a method for constructing the Minimum Euclidean or Manhattan Distance Cluster-based (MDC) deterministic compressed sensing matrix is given. Moreover, the Restricted Isometry Property of the MDC matrix is supported. Third, three groups of real neural signals are used for validation. Six different random or deterministic sensing matrices under diverse reconstruction algorithms are used for the simulation. From the simulation results, it can be demonstrated that the MDC matrix can largely compress neural signals and also have a small reconstruction error. For a six-thousand-point signal, the compression rate can be up to 98%, whereas the reconstruction error is less than 0.1. In addition, from the simulation results, the MDC matrix is optimal for a signal that has an extended length. Finally, the MDC matrix can be constructed by zeros and ones; additionally, it has a simple construction structure that is highly practicable for the design of an implantable neural recording device.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... The UWB monitoring of life signs has been widely studied in the literature [9][10][11][12][13][14] as an alternative method to Doppler radar systems [15,16]. The measurements of life signs in different conditions have been widely studied by researchers [17][18][19]. ...
... To acquire the frequency of human respiratory signals, the accumulated spectrums are considered, which are given by: (10) Usually, the maximal value in Equation (10) is defined as the frequency of human respiratory signals. ...
Article
Full-text available
This paper analyzes and discusses the capability of human being detection using impulse ultra-wideband (UWB) radar with an improved detection algorithm. The multiple automatic gain control (AGC) technique is employed to enhance the amplitudes of human respiratory signals. Two filters with seven values averaged are used to further improve the signal-to-noise ratio (SNR) of the human respiratory signals. The maximum slope and standard deviation are used for analyzing the characteristics of the received pulses, which can provide two distance estimates for human being detection. Most importantly, based on the two distance estimates, we can accurately judge whether there are human beings in the detection environments or not. The data size can be reduced based on the defined interested region, which can improve the operation efficiency of the radar system for human being detection. The developed algorithm provides excellent performance regarding human being detection, which is validated through comparison with several well-known algorithms.
... The corresponding researches on the detection and monitoring of living persons using UWB bio-radars have been considered widely in the literatures (Sana et al. 2014;Li and Sawan 2015;Kranjec et al. 2014;Nguyen et al. 2013Nguyen et al. , 2014Khan et al. 2014a, b), which can better replace the Doppler radars (Ascione et al. 2013;Liang et al. 2018a, b, c). UWB bio-radars have been applied widely in many scenarios (Lazaro et al. 2014;Yan et al. 2016;Li et al. 2013). ...
Article
Full-text available
Ultra-wideband (UWB) bio-radar is a promising technology for the contactless detection and wireless monitoring of stationary living persons remotely. However the optional detection approach is fewer and the accuracy is less. Thus some other alternative methods for the detection of living persons are urgently required. Here, we investigated the detection of living persons by extracting human respiratory signals based on the machine learning under non-line of sight (NLOS) multipath environments. Extreme learning machine (ELM) was utilized to determine if one person is present in the detection environment or not based on the distributions of vital signs acquired from the statistics fingerprint analysis (SFA). Then, the life sign points of interest (LSPOI) can be defined to measure the frequency of human cardiopulmonary movement when the living persons in present in the environments. Meanwhile the amount of data can be reduced and the system efficiency can be enhanced which plays a key role in human detection. The detection results indicate the developed algorithm outperform the known references.
Article
This paper concerns a wireless multichannel neural recording system using a compressed sensing technique to compress the recorded data. We put forth a single and a multichannel system applying a Minimum Euclidean or Manhattan Distance Cluster-based (MDC) deterministic compressed sensing matrix. The single-channel signal processing system is composed of spike detection and data compression blocks. For the construction of the MDC matrix, the distance σ is an important parameter, which can take a value of 4 or 5. In addition, the sharing strategy is used to construct a multichannel system, and we analyze the influence of the number of the channels; scan rate on the reconstruction error, compression rate and power consumption; the influence of the signal-to-noise ratio; and reconstruction performance on neural signals. Based on the results, a 256-channel digital signal processing system, implemented in a 130-nm CMOS process, is proposed. This system has power consumption per channel of 12.5 μW and silicon area per channel of 0.03 mm2, and provides data reduction of around 90% while enabling accurate reconstruction of the original signals.
Article
Full-text available
We present a mathematical connection between channel coding and compressed sensing. In particular, we link, on the one hand, channel coding linear programming decoding (CC-LPD), which is a well-known relaxation of maximum-likelihood channel decoding for binary linear codes, and, on the other hand, compressed sensing linear programming decoding (CS-LPD), also known as basis pursuit, which is a widely used linear programming relaxation for the problem of finding the sparsest solution of an underdetermined system of linear equations. More specifically, we establish a tight connection between CS-LPD based on a zero-one measurement matrix over the reals and CC-LPD of the binary linear channel code that is obtained by viewing this measurement matrix as a binary parity-check matrix. This connection allows the translation of performance guarantees from one setup to the other. The main message of this paper is that parity-check matrices of “good” channel codes can be used as provably “good” measurement matrices under basis pursuit. In particular, we provide the first deterministic construction of compressed sensing measurement matrices with an order-optimal number of rows using high-girth low-density parity-check codes constructed by Gallager.
Article
Full-text available
A 100-channel fully implantable wireless broadband neural recording system was developed. It features 100 parallel broadband (0.1 Hz-7.8 kHz) neural recording channels, a medical grade 200 mAh Li-ion battery recharged inductively at 150 kHz , and data telemetry using 3.2 GHz to 3.8 GHz FSK modulated wireless link for 48 Mbps Manchester encoded data. All active electronics are hermetically sealed in a titanium enclosure with a sapphire window for electromagnetic transparency. A custom, high-density configuration of 100 individual hermetic feedthrough pins enable connection to an intracortical neural recording microelectrode array. A 100 MHz bandwidth custom receiver was built to remotely receive the FSK signal and achieved -77.7 dBm sensitivity with 10(-8) BER at 48 Mbps data rate. ESD testing on all the electronic inputs and outputs has proven that the implantable device satisfies the HBM Class-1B ESD Standard. In addition, the evaluation of the worst-case charge density delivered to the tissue from each I/O pin verifies the patient safety of the device in the event of failure. Finally, the functionality and reliability of the complete device has been tested on-bench and further validated chronically in ongoing freely moving swine and monkey animal trials for more than one year to date.
Article
Full-text available
Compressed sensing is a sampling technique which provides a fundamentally new approach to data acquisition. Comparing with traditional methods, compressed sensing makes full use of sparsity so that a sparse signal can be reconstructed from very few measurements. A central problem in compressed sensing is the construction of sensing matrices. While random sensing matrices have been studied intensively, only a few deterministic constructions are known. Inspired by algebraic geometry codes, we introduce a new deterministic construction via algebraic curves over finite fields, which is a natural generalization of DeVore's construction using polynomials over finite fields. The diversity of algebraic curves provides numerous choices for sensing matrices. By choosing appropriate curves, we are able to construct binary sensing matrices which are superior to Devore's ones. We hope this connection between algebraic geometry and compressed sensing will provide a new point of view and stimulate further research in both areas.
Article
Full-text available
This work introduces the use of compressed sensing (CS) algorithms for data compression in wireless sensors to address the energy and telemetry bandwidth constraints common to wireless sensor nodes. Circuit models of both analog and digital implementations of the CS system are presented that enable analysis of the power/performance costs associated with the design space for any potential CS application, including analog-to-information converters (AIC). Results of the analysis show that a digital implementation is significantly more energy-efficient for the wireless sensor space where signals require high gain and medium to high resolutions. The resulting circuit architecture is implemented in a 90 nm CMOS process. Measured power results correlate well with the circuit models, and the test system demonstrates continuous, on-the-fly data processing, resulting in more than an order of magnitude compression for electroencephalography (EEG) signals while consuming only 1.9 $\mu$W at 0.6 V for sub-20 kS/s sampling rates. The design and measurement of the proposed architecture is presented in the context of medical sensors, however the tools and insights are generally applicable to any sparse data acquisition.
Conference Paper
Full-text available
Binary 0-1 measurement matrices, especially those from coding theory, were introduced to compressed sensing (CS) recently. Good measurement matrices with preferred properties, e.g., the restricted isometry property (RIP) and nullspace property (NSP), have no known general ways to be efficiently checked. Khajehnejad \emph{et al.} made use of \emph{girth} to certify the good performances of sparse binary measurement matrices. In this paper, we examine the performance of binary measurement matrices with uniform column weight and arbitrary girth under basis pursuit. Explicit sufficient conditions of exact reconstruction %only including $\gamma$ and $g$ are obtained, which improve the previous results derived from RIP for any girth $g$ and results from NSP when $g/2$ is odd. Moreover, we derive explicit $l_1/l_1$, $l_2/l_1$ and $l_\infty/l_1$ sparse approximation guarantees. These results further show that large girth has positive impacts on the performance of binary measurement matrices under basis pursuit, and the binary parity-check matrices of good LDPC codes are important candidates of measurement matrices.
Article
Full-text available
Compressed Sensing is a new paradigm for acquiring the compressible signals that arise in many applications. These signals can be approximated using an amount of informa-tion much smaller than the nominal dimension of the signal. Traditional approaches acquire the entire signal and process it to extract the information. The new approach acquires a small number of nonadaptive linear measurements of the signal and uses sophisticated algorithms to determine its information content. Emerging technologies can compute these general linear measurements of a signal at unit cost per measurement. This paper exhibits a randomized measurement ensem-ble and a signal reconstruction algorithm that satisfy four requirements: 1. The measurement ensemble succeeds for all signals, with high probability over the random choices in its construction. 2. The number of measurements of the signal is optimal, except for a factor polylogarithmic in the signal length. 3. The running time of the algorithm is polynomial in the amount of information in the signal and polylogarith-mic in the signal length. 4. The recovery algorithm offers the strongest possible type of error guarantee. Moreover, it is a fully polyno-mial approximation scheme with respect to this type of error bound. Emerging applications demand this level of performance. Yet no other algorithm in the literature simultaneously achieves all four of these desiderata.
Article
Full-text available
Sparse signal expansions represent or approximate a signal using a small number of elements from a large collection of elementary waveforms. Finding the optimal sparse expansion is known to be NP hard in general and non-optimal strategies such as Matching Pursuit, Orthogonal Matching Pursuit, Basis Pursuit and Basis Pursuit De-noising are often called upon. These methods show good performance in practical situations, however, they do not operate on the ℓ 0 penalised cost functions that are often at the heart of the problem. In this paper we study two iterative algorithms that are minimising the cost functions of interest. Furthermore, each iteration of these strategies has computational complexity similar to a Matching Pursuit iteration, making the methods applicable to many real world problems. However, the optimisation problem is non-convex and the strategies are only guaranteed to find local solutions, so good initialisation becomes paramount. We here study two approaches. The first approach uses the proposed algorithms to refine the solutions found with other methods, replacing the typically used conjugate gradient solver. The second strategy adapts the algorithms and we show on one example that this adaptation can be used to achieve results that lie between those obtained with Matching Pursuit and those found with Orthogonal Matching Pursuit, while retaining the computational complexity of the Matching Pursuit algorithm.
Article
Full-text available
Fetal ECG (FECG) telemonitoring is an important branch in telemedicine. The design of a telemonitoring system via a wireless body-area network with low energy consumption for ambulatory use is highly desirable. As an emerging technique, compressed sensing (CS) shows great promise in compressing/reconstructing data with low energy consumption. However, due to some specific characteristics of raw FECG recordings such as non-sparsity and strong noise contamination, current CS algorithms generally fail in this application. This work proposes to use the block sparse Bayesian learning (BSBL) framework to compress/reconstruct non-sparse raw FECG recordings. Experimental results show that the framework can reconstruct the raw recordings with high quality. Especially, the reconstruction does not destroy the interdependence relation among the multichannel recordings. This ensures that the independent component analysis decomposition of the reconstructed recordings has high fidelity. Furthermore, the framework allows the use of a sparse binary sensing matrix with much fewer nonzero entries to compress recordings. Particularly, each column of the matrix can contain only two nonzero entries. This shows the framework, compared to other algorithms such as current CS algorithms and wavelet algorithms, can greatly reduce code execution in CPU in the data compression stage.
Article
Full-text available
This paper reports a 128-channel neural recording integrated circuit (IC) with on-the-fly spike feature extraction and wireless telemetry. The chip consists of eight 16-channel front-end recording blocks, spike detection and feature extraction digital signal processor (DSP), ultra wideband (UWB) transmitter, and on-chip bias generators. Each recording channel has amplifiers with programmable gain and bandwidth to accommodate different types of biological signals. An analog-to-digital converter (ADC) shared by 16 amplifiers through time-multiplexing results in a balanced trade-off between the power consumption and chip area. A nonlinear energy operator (NEO) based spike detector is implemented for identifying spikes, which are further processed by a digital frequency-shaping filter. The computationally efficient spike detection and feature extraction algorithms attribute to an auspicious DSP implementation on-chip. UWB telemetry is designed to wirelessly transfer raw data from 128 recording channels at a data rate of 90 Mbit/s. The chip is realized in 0.35 mum complementary metal-oxide-semiconductor (CMOS) process with an area of 8.8 times 7.2 mm<sup>2</sup> and consumes 6 mW by employing a sequential turn-on architecture that selectively powers off idle analog circuit blocks. The chip has been tested for electrical specifications and verified in an ex vivo biological environment.
Article
Full-text available
Wireless body sensor networks (WBSN) hold the promise to be a key enabling information and communications technology for next-generation patient-centric telecardiology or mobile cardiology solutions. Through enabling continuous remote cardiac monitoring, they have the potential to achieve improved personalization and quality of care, increased ability of prevention and early diagnosis, and enhanced patient autonomy, mobility, and safety. However, state-of-the-art WBSN-enabled ECG monitors still fall short of the required functionality, miniaturization, and energy efficiency. Among others, energy efficiency can be improved through embedded ECG compression, in order to reduce airtime over energy-hungry wireless links. In this paper, we quantify the potential of the emerging compressed sensing (CS) signal acquisition/compression paradigm for low-complexity energy-efficient ECG compression on the state-of-the-art Shimmer WBSN mote. Interestingly, our results show that CS represents a competitive alternative to state-of-the-art digital wavelet transform (DWT)-based ECG compression solutions in the context of WBSN-based ECG monitoring systems. More specifically, while expectedly exhibiting inferior compression performance than its DWT-based counterpart for a given reconstructed signal quality, its substantially lower complexity and CPU execution time enables it to ultimately outperform DWT-based ECG compression in terms of overall energy efficiency. CS-based ECG compression is accordingly shown to achieve a 37.1% extension in node lifetime relative to its DWT-based counterpart for “good” reconstruction quality.
Article
Full-text available
Compressive sensing (CS) is an alternative to Shannon/Nyquist sampling for the acquisition of sparse or compressible signals that can be well approximated by just K ¿ N elements from an N -dimensional basis. Instead of taking periodic samples, CS measures inner products with M < N random vectors and then recovers the signal via a sparsity-seeking optimization or greedy algorithm. Standard CS dictates that robust signal recovery is possible from M = O(K log(N/K)) measurements. It is possible to substantially decrease M without sacrificing robustness by leveraging more realistic signal models that go beyond simple sparsity and compressibility by including structural dependencies between the values and locations of the signal coefficients. This paper introduces a model-based CS theory that parallels the conventional theory and provides concrete guidelines on how to create model-based recovery algorithms with provable performance guarantees. A highlight is the introduction of a new class of structured compressible signals along with a new sufficient condition for robust structured compressible signal recovery that we dub the restricted amplification property, which is the natural counterpart to the restricted isometry property of conventional CS. Two examples integrate two relevant signal models-wavelet trees and block sparsity-into two state-of-the-art CS recovery algorithms and prove that they offer robust recovery from just M = O(K) measurements. Extensive numerical simulations demonstrate the validity and applicability of our new theory and algorithms.
Conference Paper
Full-text available
Spike sorting is an important processing step in various neuroscientific and clinical studies. An on-chip spike-sorting DSP must provide data-rate reduction while maintaining a power density much less than 800 μW/mm<sup>2</sup>. Most existing designs either provide only spike detection for multi-channel processing, or they provide detection and feature extraction only for a single channel. We demonstrate a chip for detection, alignment, and feature extraction simultaneously for 64 channels. Spike-sorting algorithms identified from a complexity-performance analysis are implemented on ASIC using a Matlab/Simulink-based architecture design framework. The chip has a modular architecture, which allows it to be configured to process 16, 32, 48, or 64 channels. Inactive cores are power-gated to reduce power consumption when the chip operates for less than 64 channels. The chip is implemented in a 90-nm CMOS process and has a power dissipation of 130 μW (power density of 30 μW/mm<sup>2</sup>) when processing all 64 channels. A data-rate reduction of 91.25% (11.71 Mbps to 1.02 Mbps) is achieved.
Article
Full-text available
We examine the recovery of block sparse signals and extend the framework in two important directions; one by exploiting signals' intra-block correlation and the other by generalizing signals' block structure. We propose two families of algorithms based on the framework of block sparse Bayesian learning (BSBL). One family, directly derived from the BSBL framework, requires knowledge of the block structure. Another family, derived from an expanded BSBL framework, is based on a weaker assumption on the block structure, and can be used when the block structure is completely unknown. Using these algorithms we show that exploiting intra-block correlation is very helpful in improving recovery performance. These algorithms also shed light on how to modify existing algorithms or design new ones to exploit such correlation and improve performance.
Article
Full-text available
Compressed sensing (CS) is an emerging field that has attracted considerable research interest over the past few years. Previous review articles in CS limit their scope to standard discrete-to-discrete measurement architectures using matrices of randomized nature and signal models based on standard sparsity. In recent years, CS has worked its way into several new application areas. This, in turn, necessitates a fresh look on many of the basics of CS. The random matrix measurement operator must be replaced by more structured sensing architectures that correspond to the characteristics of feasible acquisition hardware. The standard sparsity prior has to be extended to include a much richer class of signals and to encode broader data models, including continuous-time signals. In our overview, the theme is exploiting signal and measurement structure in compressive sensing. The prime focus is bridging theory and practice; that is, to pinpoint the potential of structured CS strategies to emerge from the math to the hardware. Our summary highlights new directions as well as relations to more traditional CS, with the hope of serving both as a review to practitioners wanting to join this emerging field, and as a reference for researchers that attempts to put some of the existing ideas in perspective of practical applications.
Article
Full-text available
We present benchtop and in vivo experimental results from an integrated circuit designed for wireless implantable neural recording applications. The chip, which was fabricated in a commercially available 0.6- mum 2P3M BiCMOS process, contains 100 amplifiers, a 10-bit analog-to-digital converter (ADC), 100 threshold-based spike detectors, and a 902-928 MHz frequency-shift-keying (FSK) transmitter. Neural signals from a selected amplifier are sampled by the ADC at 15.7 kSps and telemetered over the FSK wireless data link. Power, clock, and command signals are sent to the chip wirelessly over a 2.765-MHz inductive (coil-to-coil) link. The chip is capable of operating with only two off-chip components: a power/command receiving coil and a 100-nF capacitor.
Article
Full-text available
Applications such as brain-machine interfaces require hardware spike sorting in order to (1) obtain single-unit activity and (2) perform data reduction for wireless transmission of data. Such systems must be low-power, low-area, high-accuracy, automatic, and able to operate in real time. Several detection and feature extraction algorithms for spike sorting are described briefly and evaluated in terms of accuracy versus computational complexity. The nonlinear energy operator method is chosen as the optimal spike detection algorithm, being most robust over noise and relatively simple. The discrete derivatives method [1] is chosen as the optimal feature extraction method, maintaining high accuracy across SNRs with a complexity orders of magnitude less than that of traditional methods such as PCA.
Article
Full-text available
This study introduces a new method for detecting and sorting spikes from multiunit recordings. The method combines the wavelet transform, which localizes distinctive spike features, with superparamagnetic clustering, which allows automatic classification of the data without assumptions such as low variance or gaussian distributions. Moreover, an improved method for setting amplitude thresholds for spike detection is proposed. We describe several criteria for implementation that render the algorithm unsupervised and fast. The algorithm is compared to other conventional methods using several simulated data sets whose characteristics closely resemble those of in vivo recordings. For these data sets, we found that the proposed algorithm outperformed conventional methods.
Conference Paper
Full-text available
Many applications in signal processing require the efficient representation and processing of data. The traditional approach to efficient signal representation is compression. In recent years, there has been a new approach to compression at the sensing level. Compressed sensing (CS) is an emerging field which is based on the revelation that a small collection of linear projections of a sparse signal contains enough information for reconstruction. In this paper, we propose an application of compressed sensing in the field of biomedical signal processing, particularly electroencophelogram (EEG) collection and storage. A compressed sensing framework is introduced for efficient representation of multichannel, multiple trial EEG data. The proposed framework is based on the revelation that EEG signals are sparse in a Gabor frame. The sparsity of EEG signals in a Gabor frame is utilized for compressed sensing of these signals. A simultaneous orthogonal matching pursuit algorithm is shown to be effective in the joint recovery of the original multiple trail EEG signals from a small number of projections.
Article
Full-text available
Recent work in field of neuroprosthetics has demonstrated that by observing the simultaneous activity of many neurons in specific regions of the brain, it is possible to produce control signals that allow animals or humans to drive cursors or prosthetic limbs directly through thoughts. As neuroprosthetic devices transition from experimental to clinical use, there is a need for fully-implantable amplification and telemetry electronics in close proximity to the recording sites. To address these needs, we developed a prototype integrated circuit for wireless neural recording from a 100-channel microelectrode array. The design of both the system-level architecture and the individual circuits were driven by severe power constraints for small implantable devices; chronically heating tissue by only a few degrees Celsius leads to cell death. Due to the high data rate produced by 100 neural signals, the system must perform data reduction as well. We use a combination of a low-power ADC and an array of "spike detectors" to reduce the transmitted data rate while preserving critical information. The complete system receives power and commands (at 6.5 kb/s) wirelessly over a 2.64-MHz inductive link and transmits neural data back at a data rate of 330 kb/s using a fully-integrated 433-MHz FSK transmitter. The 4.7times5.9 mm<sup>2</sup> chip was fabricated in a 0.5-mum 3M2P CMOS process and consumes 13.5 mW of power. While cross-chip interference limits performance in single-chip operation, a two-chip system was used to record neural signals from a Utah Electrode Array in cat cortex and transmit the digitized signals wirelessly to a receiver
Article
The long-standing analog-to-digital conversion paradigm based on Shannon/Nyquist sampling has been challenged lately, mostly in situations such as radar and communication signal processing where signal bandwidth is so large that sampling architectures constraints are simply not manageable. Compressed sensing (CS) is a new emerging signal acquisition/compression paradigm that offers a striking alternative to traditional signal acquisition. Interestingly, by merging the sampling and compression steps, CS also removes a large part of the digital architecture and might thus considerably simplify analog-to-information (A2I) conversion devices. This so-called “analog CS,” where compression occurs directly in the analog sensor readout electronics prior to analog-to-digital conversion, could thus be of great importance for applications where bandwidth is moderate, but computationally complex, and power resources are severely constrained. In our previous work (Mamaghanian, 2011), we quantified and validated the potential of digital CS systems for real-time and energy-efficient electrocardiogram compression on resource-constrained sensing platforms. In this paper, we review the state-of-the-art implementations of CS-based signal acquisition systems and perform a complete system-level analysis for each implementation to highlight their strengths and weaknesses regarding implementation complexity, performance and power consumption. Then, we introduce the spread spectrum random modulator pre-integrator (SRMPI), which is a new design and implementation of a CS-based A2I read-out system that uses spread spectrum techniques prior to random modulation in order to produce the low rate set of digital samples. Finally, we experimentally built an SRMPI prototype to compare it with state-of-the-art CS-based signal acquisition systems, focusing on critical system design parameters and constraints, and show that this new proposed architecture offers a compelling alternativ- , in particular for low power and computationally-constrained embedded systems.
Article
We consider the fundamental problem of estimating the mean of a vector y=Xβ+z, where X is an n×p design matrix in which one can have far more variables than observations, and z is a stochastic error term—the so-called “p>n” setup. When β is sparse, or, more generally, when there is a sparse subset of covariates providing a close approximation to the unknown mean vector, we ask whether or not it is possible to accurately estimate Xβ using a computationally tractable algorithm. We show that, in a surprisingly wide range of situations, the lasso happens to nearly select the best subset of variables. Quantitatively speaking, we prove that solving a simple quadratic program achieves a squared error within a logarithmic factor of the ideal mean squared error that one would achieve with an oracle supplying perfect information about which variables should and should not be included in the model. Interestingly, our results describe the average performance of the lasso; that is, the performance one can expect in an vast majority of cases where Xβ is a sparse or nearly sparse superposition of variables, but not in all cases. Our results are nonasymptotic and widely applicable, since they simply require that pairs of predictor variables are not too collinear.
Article
Wireless telemonitoring of physiological signals is an important topic in eHealth. In order to reduceon-chip energy consumption and extend sensor life, recorded signals are usually compressed beforetransmission. In this paper, we adopt compressed sensing (CS) as a low-power compression framework,and propose a fast block sparse Bayesian learning (BSBL) algorithm to reconstruct original signals. Exper-iments on real-world fetal ECG signals and epilepsy EEG signals showed that the proposed algorithmhas good balance between speed and data reconstruction fidelity when compared to state-of-the-art CSalgorithms. Further, we implemented the CS-based compression procedure and a low-power compres-sion procedure based on a wavelet transform in field programmable gate array (FPGA), showing that theCS-based compression can largely save energy and other on-chip computing resources.
Article
We give a simple technique for verifying the Restricted Isometry Property (as introduced by Candès and Tao) for random matrices that underlies Compressed Sensing. Our approach has two main ingredients: (i) concentration inequalities for random inner products that have recently provided algorithmically simple proofs of the Johnson–Lindenstrauss lemma; and (ii) covering numbers for finite-dimensional balls in Euclidean space. This leads to an elementary proof of the Restricted Isometry Property and brings out connections between Compressed Sensing and the Johnson–Lindenstrauss lemma. As a result, we obtain simple and direct proofs of Kashin’s theorems on widths of finite balls in Euclidean space (and their improvements due to Gluskin) and proofs of the existence of optimal Compressed Sensing measurement matrices. In the process, we also prove that these measurements have a certain universality with respect to the sparsity-inducing basis.
Conference Paper
A design of small, low-power, low-data rate, wireless 32-channel neural recording system for small animal head-stage is presented. A neural pre-amplifier has low-input-referred-noise of 1.95 muV<sub>rms</sub> and consumes 53.6 muW. To enable digital telemetry with optimized bandwidth under size and power constraint for small-animal headstage, we propose to separately record spikes and local-field potentials. An adaptive spike detector using absolute value algorithm accompanied with 7th-order all-pass delay filter provides accurate on-chip acquisition of spike waveform in duration of 2 ms. A low-power 10-bit and 5-bit resolution A/D converters running at 22 Ksamples/s for active spikes and 200 samples/s for local field potential, respectively, can be integrated with the proposed system. Using adaptive bandwidth control, we achieve reduction of data-rate up to seven times which provides compatibility to 1 Mbps ultra low power Bluetooth technology. Total power consumption of single channel excluding ADCs is 109.58 muW in 3.3 V power supply.
Article
In this paper, we establish the connection between the Orthogonal Optical Codes (OOC) and binary compressed sensing matrices. We also introduce deterministic bipolar m × n RIP fulfilling ±1 matrices of order k such that m ≤ O ( k (log<sub>2</sub> n )<sup>( log</sup><sub>2</sub><sup>k)/( ln log</sup><sub>2</sub><sup>k)</sup>). The columns of these matrices are binary BCH code vectors where the zeros are replaced by -1. Since the RIP is established by means of coherence, the simple greedy algorithms such as Matching Pursuit are able to recover the sparse solution from the noiseless samples. Due to the cyclic property of the BCH codes, we show that the FFT algorithm can be employed in the reconstruction methods to considerably reduce the computational complexity. In addition, we combine the binary and bipolar matrices to form ternary sensing matrices ({0,1,-1} elements) that satisfy the RIP condition.
Article
Compressed Sensing aims to capture attributes of k-sparse signals using very few measurements. In the standard compressed sensing paradigm, the N ?? C measurement matrix ?? is required to act as a near isometry on the set of all k-sparse signals (restricted isometry property or RIP). Although it is known that certain probabilistic processes generate N ?? C matrices that satisfy RIP with high probability, there is no practical algorithm for verifying whether a given sensing matrix ?? has this property, crucial for the feasibility of the standard recovery algorithms. In contrast, this paper provides simple criteria that guarantee that a deterministic sensing matrix satisfying these criteria acts as a near isometry on an overwhelming majority of k-sparse signals; in particular, most such signals have a unique representation in the measurement domain. Probability still plays a critical role, but it enters the signal model rather than the construction of the sensing matrix. An essential element in our construction is that we require the columns of the sensing matrix to form a group under pointwise multiplication. The construction allows recovery methods for which the expected performance is sub-linear in C, and only quadratic in N, as compared to the super-linear complexity in C of the Basis Pursuit or Matching Pursuit algorithms; the focus on expected performance is more typical of mainstream signal processing than the worst case analysis that prevails in standard compressed sensing. Our framework encompasses many families of deterministic sensing matrices, including those formed from discrete chirps, Delsarte-Goethals codes, and extended BCH codes.
Article
This paper proposes a best basis extension of compressed sensing recovery. Instead of regularizing the compressed sensing inverse problem with a sparsity prior in a fixed basis, our framework makes use of sparsity in a tree-structured dictionary of orthogonal bases. A new iterative thresholding algorithm performs both the recovery of the signal and the estimation of the best basis. The resulting reconstruction from compressive measurements optimizes the basis to the structure of the sensed signal. Adaptivity is crucial to capture the regularity of complex natural signals. Numerical experiments on sounds and geometrical images indeed show that this best basis search improves the recovery with respect to fixed sparsity priors.
Article
Compressive sampling offers a new paradigm for acquiring signals that are compressible with respect to an orthonormal basis. The major algorithmic challenge in compressive sampling is to approximate a compressible signal from noisy samples. This paper describes a new iterative recovery algorithm called CoSaMP that delivers the same guarantees as the best optimization-based approaches. Moreover, this algorithm offers rigorous bounds on computational cost and storage. It is likely to be extremely efficient for practical problems because it requires only matrix–vector multiplies with the sampling matrix. For compressible signals, the running time is just O(Nlog2N), where N is the length of the signal.
Article
There is an enduring quest for technologies that provide – temporally and spatially – highly resolved information on electric neuronal or cardiac activity in functional tissues or cell cultures. Here, we present a planar high-density, low-noise microelectrode system realized in microelectronics technology that features 11,011 microelectrodes (3,150 electrodes per mm2), 126 of which can be arbitrarily selected and can, via a reconfigurable routing scheme, be connected to on-chip recording and stimulation circuits. This device enables long-term extracellular electrical-activity recordings at subcellular spatial resolution and microsecond temporal resolution to capture the entire dynamics of the cellular electrical signals. To illustrate the device performance, extracellular potentials of Purkinje cells (PCs) in acute slices of the cerebellum have been analyzed. A detailed and comprehensive picture of the distribution and dynamics of action potentials (APs) in the somatic and dendritic regions of a single cell was obtained from the recordings by applying spike sorting and spike-triggered averaging methods to the collected data. An analysis of the measured local current densities revealed a reproducible sink/source pattern within a single cell during an AP. The experimental data substantiated compartmental models and can be used to extend those models to better understand extracellular single-cell potential patterns and their contributions to the population activity. The presented devices can be conveniently applied to a broad variety of biological preparations, i.e., neural or cardiac tissues, slices, or cell cultures can be grown or placed directly atop of the chips for fundamental mechanistic or pharmacological studies.
Article
Compressed sensing is a novel technique to acquire sparse signals with few measurements. Normally, compressed sensing uses random projections as measurements. Here we design deterministic measurements and an algorithm to accomplish signal recovery with computational efficiency. A measurement matrix is designed with chirp sequences forming the columns. Chirps are used since an efficient method using FFTs can recover the parameters of a small superposition. We show that this type of matrix is valid as compressed sensing measurements. This is done by bounding the eigenvalues of sub-matrices, as well as an empirical comparison with random projections. Further, by implementing our algorithm, simulations show successful recovery of signals with sparsity levels similar to those possible by matching pursuit with random measurements. For sufficiently sparse signals, our algorithm recovers the signal with computational complexity O(KlogK) for K measurements. This is a significant improvement over existing algorithms.
Article
Microelectrode arrays (MEAs) offer a powerful tool to both record activity and deliver electrical microstimulations to neural networks either in vitro or in vivo. Microelectronics microfabrication technologies now allow building high-density MEAs containing several hundreds of microelectrodes. However, dense arrays of 3D micro-needle electrodes, providing closer contact with the neural tissue than planar electrodes, are not achievable using conventional isotropic etching processes. Moreover, increasing the number of electrodes using conventional electronics is difficult to achieve into compact devices addressing all channels independently for simultaneous recording and stimulation. Here, we present a full modular and versatile 256-channel MEA system based on integrated electronics. First, transparent high-density arrays of 3D-shaped microelectrodes were realized by deep reactive ion etching techniques of a silicon substrate reported on glass. This approach allowed achieving high electrode aspect ratios, and different shapes of tip electrodes. Next, we developed a dedicated analog 64-channel Application Specific Integrated Circuit (ASIC) including one amplification stage and one current generator per channel, and analog output multiplexing. A full modular system, called BIOMEA, has been designed, allowing connecting different types of MEAs (64, 128, or 256 electrodes) to different numbers of ASICs for simultaneous recording and/or stimulation on all channels. Finally, this system has been validated experimentally by recording and electrically eliciting low-amplitude spontaneous rhythmic activity (both LFPs and spikes) in the developing mouse CNS. The availability of high-density MEA systems with integrated electronics will offer new possibilities for both in vitro and in vivo studies of large neural networks.
Article
We present a low-power complementary metal-oxide semiconductor (CMOS) analog integrated biopotential detector intended for neural recording in wireless multichannel implants. The proposed detector can achieve accurate automatic discrimination of action potential (APs) from the background activity by means of an energy-based preprocessor and a linear delay element. This strategy improves detected waveforms integrity and prompts for better performance in neural prostheses. The delay element is implemented with a low-power continuous-time filter using a ninth-order equiripple allpass transfer function. All circuit building blocks use subthreshold OTAs employing dedicated circuit techniques for achieving ultra low-power and high dynamic range. The proposed circuit function in the submicrowatt range as the implemented CMOS 0.18- microm chip dissipates 780 nW, and it features a size of 0.07 mm(2). So it is suitable for massive integration in a multichannel device with modest overhead. The fabricated detector succeeds to automatically detect APs from underlying background activity. Testbench validation results obtained with synthetic neural waveforms are presented.
Article
Microelectrode recording during deep brain stimulation surgery improves the likelihood of successful target localization and enables the electrophysiological characterization of human neural structures. Many clinical recording systems do not support the ability to capture research-quality recordings. Established clinical centers already using such equipment may be prevented from acquiring human intracranial data because of the need to completely change recording systems to obtain research-quality recordings. This technical note describes the novel design and implementation of a recording system that significantly improves research capabilities without disrupting the existing clinical setup. This design introduces a second recording system (including pre-amplifier, differential amplifier, analog-to-digital converter, and computer with analysis software) that divides the microelectrode signal into two independent streams. This design preserves the existing intraoperative recording setup, but significantly improves research-level recording, data storage, and analysis capabilities. We provide the first description of such a system using components that are all commercially available and relatively inexpensive. This approach presents an appealing alternative to the purchase of an entirely new system for surgical teams that already perform intraoperative recordings to assist in stereotactic target localization, yet wish to expand their neurophysiological recording capabilities.
Conference Paper
This paper proposes a deterministic compressed sensing matrix that comes by design with a very fast reconstruction algorithm, in the sense that its complexity depends only on the number of measurements n and not on the signal dimension N. The matrix construction is based on the second order Reed-Muller codes and associated functions. This matrix does not have RIP uniformly with respect to all kappa-sparse vectors, but it acts as a near isometry on kappa-sparse vectors with very high probability.
Article
This paper considers a natural error correcting problem with real valued input/output. We wish to recover an input vector f∈R<sup>n</sup> from corrupted measurements y=Af+e. Here, A is an m by n (coding) matrix and e is an arbitrary and unknown vector of errors. Is it possible to recover f exactly from the data y? We prove that under suitable conditions on the coding matrix A, the input f is the unique solution to the ℓ<sub>1</sub>-minimization problem (||x||<sub>ℓ1</sub>:=Σ<sub>i</sub>|x<sub>i</sub>|) min(g∈R<sup>n</sup>) ||y - Ag||<sub>ℓ1</sub> provided that the support of the vector of errors is not too large, ||e||<sub>ℓ0</sub>:=|{i:e<sub>i</sub> ≠ 0}|≤ρ·m for some ρ>0. In short, f can be recovered exactly by solving a simple convex optimization problem (which one can recast as a linear program). In addition, numerical experiments suggest that this recovery procedure works unreasonably well; f is recovered exactly even in situations where a significant fraction of the output is corrupted. This work is related to the problem of finding sparse solutions to vastly underdetermined systems of linear equations. There are also significant connections with the problem of recovering signals from highly incomplete measurements. In fact, the results introduced in this paper improve on our earlier work. Finally, underlying the success of ℓ<sub>1</sub> is a crucial property we call the uniform uncertainty principle that we shall describe in detail.