Optimization is everywhere! Nearly all engineering designs rely on minimal or maximal solutions in the presence of constraints. Engineering efforts also involve optimization methods when solution is not explicit. In many cases the implementation of an idea requires engineering choices to be made, that reflect the need to optimize/ minimization in terms of design choices, software implementation or choices at the algorithmic or architectural levels. The papers in this special issue reflect the diverse ways in which optimization enters the design arena. The papers listed later reflect the use of numerical optimization based approaches, while the earlier ones show how optimization might enter into more complex design choices, such as into problem representations or at the architecture level.
The first paper in this special issue entitled “Efficient CNN Accelerator on FPGA” by Kala S and Nalesh S, deals with the FPGA Realization of Convolutional Neural Networks (CNNs). CNNs are of major interest in deep learning, but they involve compute and memory intensive operations, and their hardware realization is a challenge. Kala and Nalesh present an efficient ResNet-18 CNN accelerator based on a block Winograd-GEMM architecture. This manuscript is an extended version of work presented at the 32nd International Conference on VLSI Design and the 18th International Conference on Embedded Systems (VLSID 2019).
The second paper in the issue entitled “A Low-Power VLSI Implementation of RFIR Filter Design using Radix-2 Algorithm with LCSLA”, by Kasarla Satish Reddy and Hosahally Narayangowda Suresh deals with an efficient Reconfigurable FIR filter design with a focus on minimizing area and power. The paper relies on a Radix-2 algorithm and focusses on the multiplier and adder design.
The next paper by Majid Alotaibi entitled “A Double-Balanced UP-Conversion Mixer Using 90 nm CMOS Technology for 60 GHz Spectrum Frequency”, addresses the challenging problem of designing a 60 Ghz active mixer. Such a design needs to trade-off multiple objectives such as linearity, conversion gain, and rejection of LO components at the RF ports. Engineering such a design requires addressing multiple objectives and multiple constraints.
The following paper, entitled “Optimized Structural Compressed Sensing Matrices for Speech Compression” by Yuvraj V Parkale and Sanjay L Nalbalwar employs an optimization method based on structural sensing matrices and Grassmannian frame theory, for compressive sensing. They investigate structured sensing matrices and their optimization for optimizing coherence.
The work reported in the paper entitled “Energy-Efficient Resource Allocation in Multicell Large-Scale Distributed Antenna System with Imperfect CSI” by Meha Mahajan and Wonsik Yoon formulates and solves a non-convex optimization problem to maximize energy efficiency in the context of resource allocation algorithm in the downlink of a large-scale distributed antenna system.
The next paper in this special issue, entitled “Design Optimization for Programmable Microfluidic Devices Integrating Contamination Removal and Capacity-Wastage-Aware Washing”, is by Piyali Datta, Arpan Chakraborty and Rajat Kumar Pal. The paper discusses containment minimization design, as well as a wash optimization model. The authors use simulated annealing to find good solutions to a NP complete optimization task.
In the work by Nilotpal Chakraborty, Arijit Mondal, and Samrat Mondal entitled “Scheduling Interdependent Smart Appliances with Mixed-Preemption Policy for Peak Load Minimization”, the authors study and analyze the problem of peak load optimization when deferrable preemptive and non-preemptive loads are present. Challenges regarding practical circumstances with different scheduling problems, NP-hard problems, other constraints are presented. The authors then give a mathematical formulation and an algorithm for its solution. This paper is an extended version of work presented at the 32nd International Conference on VLSI Design and the 18th International Conference on Embedded Systems (VLSID 2019).
The next paper “Efficiency Enhancement in a Wireless Power Transmission System using Parasitic Components”, by Kim Ho Yeap, Agnes Ruey Chyi Cheah, Kee Choon Yeong, Takefumi Hiraguri, Shunya Yokoyama, Isamu Shitara, Kazuhiro Hirasawa, and Ryoji Wakabayashi, applies optimization to the design of a wireless power transmission (WPT) system. The authors use simulated annealing to optimize the positions of receiving antennae.
Khelil Fertas, Soufiane Tebache, Farid Ghanem, Smail Tedjini, and Rabia Aksas present the use of Genetic Algorithms to optimize a multiband patch antenna, in their paper entitled “Non-conventional Multiband Patch Antenna Design with Filtering Aspect Based on Genetic Algorithm”. Starting from a random matrix of squared pixel shapes, optimization is used to arrive at a final design that meets design specifications.
The authors Ye-Shun Shen, Fang-Biau Ueng, and Chin-Yang Kung in their paper entitled “Novel Detectors for Massive MU-MIMO Communications” present the design of a low-complexity turbo receiver with joint multiuser detector, channel estimator and turbo equalizer for MU-MMIMO SC-FDMA. Although optimization is not explicitly stated, an estimation problem is solved as part of their overall approach.
In the concluding paper entitled “Performance and Energy Conservation of 3GPP IFOM Protocol for Dual Connectivity in Heterogeneous LTE-WLAN Network”, by Shubhada Gadgil, Shashi Ranjan, and Abhay Karandikar, an interesting problem in the context of IP Flow Mobility (IFOM), that is relevant in the context of 5G networks is discussed. They examine throughput maximization in heterogeneous networks, resulting in energy cost reduction, while keeping latency low and satisfying Quality of Service (QoS) requirements. Their work shows that a system model is often the significant component of the design effort.
The papers selected for this Special Issue on “Design Optimization” include some chosen from 100 submissions recommended for publication in the IETE Journal of Research, with some being expanded versions of manuscripts submitted to the 32nd International Conference on VLSI Design and Embedded Systems held in New Delhi in 2018. Thanks are due to the authors whose papers have been selectively chosen for this special issue.
It may be interesting to explore how the problems of the kind represented by the first group of papers may be attacked by methods employed in the second set. Perhaps such an effort could be the focus of a future issue.
I hope that you find the papers in this special issue stimulating and valuable. Happy reading!
Additional information
Notes on contributors
Jayadeva
Jayadeva is an Institute Chair Professor at IIT Delhi and Head of the Electrical Engineering Department. His notable work includes the Twin Support Vector Machine that has over 1100 Google Scholar citations; it is the subject of survey papers in AI Review (2012), Neural Computing and Applications (2014), Annals of Data Science (2014), Egyptian Informatics Journal (2015), Neurocomputing (2018), and Appl. Soft Computing (2019). He received best paper awards for two conference papers and one journal paper; one of his papers was listed on Neurocomputing’s Top 25 Hotlist. His team was amongst the first around the world to design and realize a SVM based A/D converter on chip (2006). Another 2015 chip is, to the best of knowledge, the first ant colony optimization algorithm on silicon. His co-authored rare cell detection algorithm in Nature Communications (2018) discovered a new cell subtype. He was a program chair for VLSI Design 2019, and has been an invited speaker at multiple fora. His patents include ones on A/D conversion, a method for assessing pronunciation abilities, and optimization free chromatography. He is a co-author of the books “Numerical Optimization with Applications”, and “Twin Support Vector Machines: Models, Extensions, and Applications” (Springer, 2017). He is a fellow of the INAE and the IETE. E-mail: [email protected]