skip to main content

Title: Training for Chanel Estimation in Nonlinear Multi-Antenna Transceivers
Recent efforts to obtain high data rates in wireless systems have focused on what can be achieved in systems that have nonlinear or coarsely quantized transceiver architectures. Estimating the channel in such a system is challenging because the nonlinearities distort the channel estimation process. It is therefore of interest to determine how much training is needed to estimate the channel sufficiently well so that the channel estimate can be used during data communication. We provide a way to determine how much training is needed by deriving a lower bound on the achievable rate in a training-based scheme that can be computed and analyzed even when the number of antennas is very large. This lower bound can be tight, especially at high SNR. One conclusion is that the optimal number of training symbols may paradoxically be smaller than the number of transmitters for systems with coarselyquantized transceivers. We show how the training time can be strongly dependent on the number of receivers, and give an example where doubling the number of receivers reduces the training time by about 37 percent.
Authors:
Award ID(s):
1731056
Publication Date:
NSF-PAR ID:
10104226
Journal Name:
Information Theory and Applications
Page Range or eLocation-ID:
1-11
Sponsoring Org:
National Science Foundation
More Like this
  1. Downlink channel estimation in massive MIMO systems is well known to generate a large overhead in frequency division duplex (FDD) mode as the amount of training generally scales with the number of transmit antennas. Using instead an extrapolation of the channel from the measured uplink estimates to the downlink frequency band completely removes this overhead. In this paper, we investigate the theoretical limits of channel extrapolation in frequency. We highlight the advantage of basing the extrapolation on high-resolution channel estimation. A lower bound (LB) on the mean squared error (MSE) of the extrapolated channel is derived. A simplified LB is also proposed, giving physical intuition on the SNR gain and extrapolation range that can be expected in practice. The validity of the simplified LB relies on the assumption that the paths are well separated. The SNR gain then linearly improves with the number of receive antennas while the extrapolation performance penalty quadratically scales with the ratio of the frequency and the training bandwidth. The theoretical LB is numerically evaluated using a 3GPP channel model and we show that the LB can be reached by practical high-resolution parameter extraction algorithms. Our results show that there are strong limitations on the extrapolationmore »range than can be expected in SISO systems while much more promising results can be obtained in the multiple-antenna setting as the paths can be more easily separated in the delay-angle domain.« less
  2. Abstract Background

    The topology of metabolic networks is both well-studied and remarkably well-conserved across many species. The regulation of these networks, however, is much more poorly characterized, though it is known to be divergent across organisms—two characteristics that make it difficult to model metabolic networks accurately. While many computational methods have been built to unravel transcriptional regulation, there have been few approaches developed for systems-scale analysis and study of metabolic regulation. Here, we present a stepwise machine learning framework that applies established algorithms to identify regulatory interactions in metabolic systems based on metabolic data: stepwise classification of unknown regulation, or SCOUR.

    Results

    We evaluated our framework on both noiseless and noisy data, using several models of varying sizes and topologies to show that our approach is generalizable. We found that, when testing on data under the most realistic conditions (low sampling frequency and high noise), SCOUR could identify reaction fluxes controlled only by the concentration of a single metabolite (its primary substrate) with high accuracy. The positive predictive value (PPV) for identifying reactions controlled by the concentration of two metabolites ranged from 32 to 88% for noiseless data, 9.2 to 49% for either low sampling frequency/low noise or high sampling frequency/high noisemore »data, and 6.6–27% for low sampling frequency/high noise data, with results typically sufficiently high for lab validation to be a practical endeavor. While the PPVs for reactions controlled by three metabolites were lower, they were still in most cases significantly better than random classification.

    Conclusions

    SCOUR uses a novel approach to synthetically generate the training data needed to identify regulators of reaction fluxes in a given metabolic system, enabling metabolomics and fluxomics data to be leveraged for regulatory structure inference. By identifying and triaging the most likely candidate regulatory interactions, SCOUR can drastically reduce the amount of time needed to identify and experimentally validate metabolic regulatory interactions. As high-throughput experimental methods for testing these interactions are further developed, SCOUR will provide critical impact in the development of predictive metabolic models in new organisms and pathways.

    « less
  3. Obeid, I. ; Selesnick, I. (Ed.)
    The Neural Engineering Data Consortium at Temple University has been providing key data resources to support the development of deep learning technology for electroencephalography (EEG) applications [1-4] since 2012. We currently have over 1,700 subscribers to our resources and have been providing data, software and documentation from our web site [5] since 2012. In this poster, we introduce additions to our resources that have been developed within the past year to facilitate software development and big data machine learning research. Major resources released in 2019 include: ● Data: The most current release of our open source EEG data is v1.2.0 of TUH EEG and includes the addition of 3,874 sessions and 1,960 patients from mid-2015 through 2016. ● Software: We have recently released a package, PyStream, that demonstrates how to correctly read an EDF file and access samples of the signal. This software demonstrates how to properly decode channels based on their labels and how to implement montages. Most existing open source packages to read EDF files do not directly address the problem of channel labels [6]. ● Documentation: We have released two documents that describe our file formats and data representations: (1) electrodes and channels [6]: describes how tomore »map channel labels to physical locations of the electrodes, and includes a description of every channel label appearing in the corpus; (2) annotation standards [7]: describes our annotation file format and how to decode the data structures used to represent the annotations. Additional significant updates to our resources include: ● NEDC TUH EEG Seizure (v1.6.0): This release includes the expansion of the training dataset from 4,597 files to 4,702. Calibration sequences have been manually annotated and added to our existing documentation. Numerous corrections were made to existing annotations based on user feedback. ● IBM TUSZ Pre-Processed Data (v1.0.0): A preprocessed version of the TUH Seizure Detection Corpus using two methods [8], both of which use an FFT sliding window approach (STFT). In the first method, FFT log magnitudes are used. In the second method, the FFT values are normalized across frequency buckets and correlation coefficients are calculated. The eigenvalues are calculated from this correlation matrix. The eigenvalues and correlation matrix's upper triangle are used to generate feature. ● NEDC TUH EEG Artifact Corpus (v1.0.0): This corpus was developed to support modeling of non-seizure signals for problems such as seizure detection. We have been using the data to build better background models. Five artifact events have been labeled: (1) eye movements (EYEM), (2) chewing (CHEW), (3) shivering (SHIV), (4) electrode pop, electrostatic artifacts, and lead artifacts (ELPP), and (5) muscle artifacts (MUSC). The data is cross-referenced to TUH EEG v1.1.0 so you can match patient numbers, sessions, etc. ● NEDC Eval EEG (v1.3.0): In this release of our standardized scoring software, the False Positive Rate (FPR) definition of the Time-Aligned Event Scoring (TAES) metric has been updated [9]. The standard definition is the number of false positives divided by the number of false positives plus the number of true negatives: #FP / (#FP + #TN). We also recently introduced the ability to download our data from an anonymous rsync server. The rsync command [10] effectively synchronizes both a remote directory and a local directory and copies the selected folder from the server to the desktop. It is available as part of most, if not all, Linux and Mac distributions (unfortunately, there is not an acceptable port of this command for Windows). To use the rsync command to download the content from our website, both a username and password are needed. An automated registration process on our website grants both. An example of a typical rsync command to access our data on our website is: rsync -auxv nedc_tuh_eeg@www.isip.piconepress.com:~/data/tuh_eeg/ Rsync is a more robust option for downloading data. We have also experimented with Google Drive and Dropbox, but these types of technology are not suitable for such large amounts of data. All of the resources described in this poster are open source and freely available at https://www.isip.piconepress.com/projects/tuh_eeg/downloads/. We will demonstrate how to access and utilize these resources during the poster presentation and collect community feedback on the most needed additions to enable significant advances in machine learning performance.« less
  4. Obeid, I. ; Selesnik, I. ; Picone, J. (Ed.)
    The Neuronix high-performance computing cluster allows us to conduct extensive machine learning experiments on big data [1]. This heterogeneous cluster uses innovative scheduling technology, Slurm [2], that manages a network of CPUs and graphics processing units (GPUs). The GPU farm consists of a variety of processors ranging from low-end consumer grade devices such as the Nvidia GTX 970 to higher-end devices such as the GeForce RTX 2080. These GPUs are essential to our research since they allow extremely compute-intensive deep learning tasks to be executed on massive data resources such as the TUH EEG Corpus [2]. We use TensorFlow [3] as the core machine learning library for our deep learning systems, and routinely employ multiple GPUs to accelerate the training process. Reproducible results are essential to machine learning research. Reproducibility in this context means the ability to replicate an existing experiment – performance metrics such as error rates should be identical and floating-point calculations should match closely. Three examples of ways we typically expect an experiment to be replicable are: (1) The same job run on the same processor should produce the same results each time it is run. (2) A job run on a CPU and GPU should producemore »identical results. (3) A job should produce comparable results if the data is presented in a different order. System optimization requires an ability to directly compare error rates for algorithms evaluated under comparable operating conditions. However, it is a difficult task to exactly reproduce the results for large, complex deep learning systems that often require more than a trillion calculations per experiment [5]. This is a fairly well-known issue and one we will explore in this poster. Researchers must be able to replicate results on a specific data set to establish the integrity of an implementation. They can then use that implementation as a baseline for comparison purposes. A lack of reproducibility makes it very difficult to debug algorithms and validate changes to the system. Equally important, since many results in deep learning research are dependent on the order in which the system is exposed to the data, the specific processors used, and even the order in which those processors are accessed, it becomes a challenging problem to compare two algorithms since each system must be individually optimized for a specific data set or processor. This is extremely time-consuming for algorithm research in which a single run often taxes a computing environment to its limits. Well-known techniques such as cross-validation [5,6] can be used to mitigate these effects, but this is also computationally expensive. These issues are further compounded by the fact that most deep learning algorithms are susceptible to the way computational noise propagates through the system. GPUs are particularly notorious for this because, in a clustered environment, it becomes more difficult to control which processors are used at various points in time. Another equally frustrating issue is that upgrades to the deep learning package, such as the transition from TensorFlow v1.9 to v1.13, can also result in large fluctuations in error rates when re-running the same experiment. Since TensorFlow is constantly updating functions to support GPU use, maintaining an historical archive of experimental results that can be used to calibrate algorithm research is quite a challenge. This makes it very difficult to optimize the system or select the best configurations. The overall impact of all of these issues described above is significant as error rates can fluctuate by as much as 25% due to these types of computational issues. Cross-validation is one technique used to mitigate this, but that is expensive since you need to do multiple runs over the data, which further taxes a computing infrastructure already running at max capacity. GPUs are preferred when training a large network since these systems train at least two orders of magnitude faster than CPUs [7]. Large-scale experiments are simply not feasible without using GPUs. However, there is a tradeoff to gain this performance. Since all our GPUs use the NVIDIA CUDA® Deep Neural Network library (cuDNN) [8], a GPU-accelerated library of primitives for deep neural networks, it adds an element of randomness into the experiment. When a GPU is used to train a network in TensorFlow, it automatically searches for a cuDNN implementation. NVIDIA’s cuDNN implementation provides algorithms that increase the performance and help the model train quicker, but they are non-deterministic algorithms [9,10]. Since our networks have many complex layers, there is no easy way to avoid this randomness. Instead of comparing each epoch, we compare the average performance of the experiment because it gives us a hint of how our model is performing per experiment, and if the changes we make are efficient. In this poster, we will discuss a variety of issues related to reproducibility and introduce ways we mitigate these effects. For example, TensorFlow uses a random number generator (RNG) which is not seeded by default. TensorFlow determines the initialization point and how certain functions execute using the RNG. The solution for this is seeding all the necessary components before training the model. This forces TensorFlow to use the same initialization point and sets how certain layers work (e.g., dropout layers). However, seeding all the RNGs will not guarantee a controlled experiment. Other variables can affect the outcome of the experiment such as training using GPUs, allowing multi-threading on CPUs, using certain layers, etc. To mitigate our problems with reproducibility, we first make sure that the data is processed in the same order during training. Therefore, we save the data from the last experiment and to make sure the newer experiment follows the same order. If we allow the data to be shuffled, it can affect the performance due to how the model was exposed to the data. We also specify the float data type to be 32-bit since Python defaults to 64-bit. We try to avoid using 64-bit precision because the numbers produced by a GPU can vary significantly depending on the GPU architecture [11-13]. Controlling precision somewhat reduces differences due to computational noise even though technically it increases the amount of computational noise. We are currently developing more advanced techniques for preserving the efficiency of our training process while also maintaining the ability to reproduce models. In our poster presentation we will demonstrate these issues using some novel visualization tools, present several examples of the extent to which these issues influence research results on electroencephalography (EEG) and digital pathology experiments and introduce new ways to manage such computational issues.« less
  5. One-bit transceivers with strongly nonlinear characteristics are being considered for wireless communication because of their low cost and low power consumption. Although each such transceiver can support only a low data rate, multiple such transceivers can be used to obtain an aggregate high data rate. An important part of many communication systems is the process of channel estimation, which is particularly challenging when the estimation process uses these transceivers. The standard analysis of estimation mean-square error versus training length that is available for linear transceivers does not apply with the nonlinearities inherent in one-bit transceivers. We analyze the training requirements in a large- scale system and show that the optimal number of training symbols strongly depends on the number of receivers, and the optimal number of training symbols can be significantly smaller than the number of transmitters. These results contrast sharply with classical results obtained with linear transceivers.