DRSR@DA-IICT
http://drsr.daiict.ac.in:80
The DSpace digital repository system captures, stores, indexes, preserves, and distributes digital research material.2021-10-18T12:15:55Z3D shape deformations : a lie group based approach
http://drsr.daiict.ac.in//handle/123456789/895
3D shape deformations : a lie group based approach
Bansal, Sumukh
3D shapes are ubiquitous in many fundamental tasks of computer graphics and geometry processing. For many applications, new shapes have to be generated from the existing ones, for which it it imperative to understand and model shape of an object and its deformation. This thesis focuses on shape deformations and its applications. Real world 3D objects undergo complex, often non-rigid deformations. One way to model such deformations is using local affine transformations. It is thus important for applications like 3D animation, to understand the structure of affine transformations and come up with robust and efficient computational tools on the set of affine transformations. With such tools, applications like interactive shape deformation and mesh interpolation can be effectively dealt with. In this thesis, an interpolation framework for affine transformations, based on a Lie group representation of a tetrahedron is proposed. The proposed framework provides a intuitive closed form interpolation in all cases in contrast to existing approaches and preserves properties like isometry, reversibility, and monotonic change of volume. The proposed Lie group representation of the tetrahedron is extended to represent triangular and tetrahedral meshes. A detailed analysis of the invariance of the representation and interpolation to some choices made, is provided in the thesis. We demonstrate the applicability of the framework for several applications like interactive shape deformation, shape interpolation, morphing, and deformation transfer. The proposed interactive shape deformation algorithm is close to being real-time, while the mesh interpolation algorithm is able to handle nonregistered meshes and large deformation cases. The interactive shape deformavi tion algorithm is amenable to data-driven methods, and we hope to explore datadriven methods using our mesh representation in near future.
2020-01-01T00:00:00ZInvestigating into a light-weight reconfigurable VLSI architecture for biomedical signal processing applications
http://drsr.daiict.ac.in//handle/123456789/892
Investigating into a light-weight reconfigurable VLSI architecture for biomedical signal processing applications
Jain, Nupur
The Body Sensor Network systems consist of signal acquisition and processing blocks along with Power Management Unit and radio transmission capabilities. The high power consumption of the radio transmission is often eliminated by adopting the on-node processing through signal processing platform with increased computation ability. Dedicated hardware accelerators optimized for operations predominantly seen in biomedical signal processing algorithms are oftenused in tandem with a microprocessor for this purpose. However, they do not support further algorithm improvements and optimizations owing to their dedicated nature. The benefits of configurability can be found in reconfigurable architectures at the cost of reconfiguration overheads. The shift-accumulate architecture developed in this thesis leverage the regularity in dominant functions in biomedical signal processing and thereby yields gate count advantages. The configurable datapath of the architecture renders multiple DSP operation emulation by means of mapping methodologies developed for efficient realization in terms of hardware utilization and memory accesses. The architecture exhibits various topologies which further supports efficient function realization. The configuration scheme of the architecture is developed which effectively consist of control word and tightly coupled data memory. The architecture is realized on a Filed Programmable Gate Array (FPGA) platform demonstrating the target function emulation and hardware results are compared with ideal outcomes. The Video Graphics Array (VGA) and Universal Asynchronous Receiver Transmitter (UART) interface controllers are developed in this work for error quantification and analysis. The architecture contains a 6 6 array of functional units having shift-accumulate as its underlying operation and has gate count of 25k and 46.9 MHz operating frequency while emulating 36-tap FIR, CORDIC, DCT, DWT, moving average, squaring and differentiation functions. Generally, biomedical signal processing functions include multiple stages consisting of noise removal, feature detection and extraction etc. The on-the-fly reconfigurability is incorporated into the architecture that leverage the low input datarates of biosignals. The architecture reconfigures dynamically while realizing different functions of the signal chain. The memory adapts to the incoming target function and supports 7 functions in its present structure. However, the architecture and memory remains scalable. Pan-Tompkins Algorithm based QRS detection realization is demonstrated on the architecture using the reconfigurability. This work offers 4 reduced area and 2.3 increase in performance with respect to the existing contemporary literatures.
2019-01-01T00:00:00ZCompressive sampling architecture for wideband communication
http://drsr.daiict.ac.in//handle/123456789/894
Compressive sampling architecture for wideband communication
Prakash, Chandra
This dissertation proposes a novel Compressive Sampling (CS) scheme for Sub-Nyquist Spectrum Sensing (SNSS) of spectrally sparse wideband signals. A novelty of our proposed SNSS scheme resides in the analog front-end. We show that it can be modeled as a sparse binary-valued measurement matrix. This has allowed us to bring to bear the proven advantages of the Low Density Parity Check (LDPC) matrices in improving the performance of the existing SNSS methods. Specifically, we show that the number of parallel SNSS channels required for a robust CS sparsity detection in our proposal is reduced compared to the existing SNSS methods. We provide new analytic (information-theoretic) lower bounds on this number and show that the LDPC-based measurement matrix is closer to this bound compared to the alternatives.The existing algorithms (such as those based on Matching Pursuit or Basis Pursuit)for CS sparsity detection are not optimal for our proposed architecture giventhe unique (sparse binary-valued) aspect of the measurement matrix. We developtwo new Belief Propagation (BP) algorithms - an Independent Probability Estimates(IPE) algorithm and a Joint Probability Estimates (JPE) algorithm - to solvethe sparsity detection problem. The performance of these algorithms is evaluatedusing Monte-Carlo simulations as well as semi-analytic approaches based onDensity Evolution and EXIT (Extrinsic Information Transfer) methods. We showthat the proposed algorithms outperform several existing algorithms (includingthe well-known Orthogonal Matching Pursuit (OMP) algorithm).Another contribution of our work is in mitigating the problem of noise enhancement (during Zero-Forcing based signal reconstruction) that affects several existing SNSS schemes (such as the Modulated Wideband Converter (MWC)). We provide analytical proofs showing this benefit and confirm the analytical results by simulation.Finally, we demonstrate the signal reconstruction in the proposed CS receiver through simulation. The Bit Error Rate (BER) performance of a QPSK system with the proposed CS receiver is simulated and the performance improvement over the MWC is demonstrated. As an extension of the developed algorithms, a framework of joint compression and denoising application is envisioned and presented with theoretical analysis.
2020-01-01T00:00:00ZOn heterogeneous distributed storage systems: bounds and code constructions
http://drsr.daiict.ac.in//handle/123456789/891
On heterogeneous distributed storage systems: bounds and code constructions
Gopal, Krishna
In Distributed Storage Systems (DSSs), usually, data is stored using encoded packets on different chunk servers. In this thesis, we have considered heterogeneous DSSs in which each node may store a different number of packets and each having different repair bandwidth. In particular, a data collector can reconstruct the file at time t using some specific nodes in the system, and for arbitrary node failure, the system can be repaired by some set of arbitrary nodes. Using min-cut bound, we investigate the fundamental trade-off between storage and repair cost for our model of heterogeneous DSS. In particular, the problem is formulated as a biobjective optimization linear programming problem. For an arbitrary DSS, it is shown that the calculated min-cut bound is tight. For a DSS with symmetric parameters, a well known class of Distributed Replication-based Simple Storage (DRESS) codes is Fractional Repetition (FR) code. In such systems, the replicas of data packets encoded by Maximum Distance Separable (MDS) code, are stored on distributed nodes. Most of the available constructions for the FR codes are based on combinatorial designs and Graph theory. In this thesis, FR codes with generalized parameters (such as replication factor of each packet are not same and storage capacity of each node are also not same) are considered, and it is called as Generalized Fractional Repetition (GFR) code. For the GFR code, we propose an elegant sequence-based approach for the construction of the GFR code called Flower codes. Further, it is shown that any GFR code is equivalent to a Flower code. The condition for the universally good GFR code is given on such sequences. For some sequences, the universally good GFR codes are explored. In general, for the GFR codes with non-uniform parameters, bounds on the GFR code rate and DSS code rate are studied. Further, we have shown that a GFR code corresponds to a hypergraph. Using the correspondence, properties and bounds of a hypergraph are directly mapped to the associated GFR code. In general, necessary and sufficient conditions for the existence of a GFR code is obtained using the correspondence. It is also shown that any GFR code associated with a linear hypergraph is universally good.
2019-01-01T00:00:00Z