Abstract
Despite the progress made in digital signal processing during the last decades, the constraints imposed by high data rate communications are becoming ever more stringent. Moreover mobile communications raised the importance of power consumption for sophisticated algorithms, such as channel equalization or decoding. The strong link existing between computational speed and power consumption suggests an investigation of signal processing with energy efficiency as a prominent design choice. In this work we revisit the topic of signal processing with analog circuits and its potential to increase the energy efficiency. Channel equalization is chosen as an application of nonlinear signal processing, and a vector equalizer based on a recurrent neural network structure is taken as an example to demonstrate what can be achieved with state of the art in VLSI design. We provide an analysis of the equalizer, including the analog circuit design, system-level simulations, and comparisons with the theoretical algorithm. First measurements of our analog VLSI circuit confirm the possibility to achieve an energy requirement of a few pJ/bit, which is an improvement factor of three to four orders of magnitude compared with today’s most energy efficient digital circuits.
Similar content being viewed by others
Notes
Microprocessors included in the comparison [Name, declared peak performance, TDP]: (1) Intel i7-3930k, 153 GFLOPS, 130 W; (2) Intel i7-3840QM, 89.6 GFLOPS, 45 W; (3) Intel i5-3570, 108.8 GFLOPS, 77 W; (4) Intel i5-3610ME, 43 GFLOPS, 35 W; (5) Intel i3-3229Y, 22.4 GFLOPS, 13 W.
References
Cauwenberghs, G. (1996). An analog VLSI recurrent neural network learning a continuous-time trajectory. IEEE Transactions on Neural Network, 2, 346–361.
CompuGreen-LLC: The green500 list (2015). http://www.green500.org [Accessed on December 2015].
Degnan, B., Marr, B., & Hasler, J. (2016). Assessing trends in performance per watt for signal processing applications. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 24(1), 58–66.
Draghici, S. (2000). Neural networks in analog hardware - design and implementation issues. International Journal of Neural Systems, 10(1), 19–42.
Engelhart, A. (2003). Vector detection techniques with moderate complexity. Ph.D. thesis, Ulm University Institute of Information Technology.
Engelhart, A., Teich, W.G., Lindner, J., Jeney, G., Imre, S., & Pap, L. (2002). A survey of multiuser/multisubchannel detection schemes based on recurrent neural networks. Wireless Communications and Mobile Computing, Special Issue on Advances in 3G Wireless Networks, 2(3), 269–284.
Haykin, S. (1994). Neural networks: A comprehensive foundation. USA: Macmillan college publishing company, Inc.
Kechriotis, G.I., & Manolakos, E.S. (1996). Hopfield neural network implementation for optimum CDMA multiuser detector. IEEE Transactions on Neural Networks, 7(1), 131–141.
Kothapalli, G. (2005). An analogue recurrent neural network for trajectory learning and other industrial applications. In 3Rd IEEE international conference on industrial informatics (INDIN) (pp. 462–466). Western Australia: Pert.
Kuroe, Y., Hashimoto, N., & Mori, T. (2002). On energy function for complex-valued neural networks and its applications. In Proc. of the 9th international conference on neural information processing ICONIP’02, (Vol. 3 pp. 1079–1083).
Lindner, J. (1999). MC-CDMA in the context of general multiuser/ multisubchannel transmission methods. European Transactions on Telecommunications, 10(4), 351–367.
Loeliger, H.A. (1999). Decoding in analog VLSI. IEEE Communications Magazine, 37(4), 99–101.
Markram, H. (2012). The human brain project - a report to the european commission. Tech. rep., The HBP-PS Consortium.
Marr, B., Degnan, B., Hasler, P., & Anderson, D. (2013). Scaling energy per operation via an asynchronous pipeline. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 21(1), 147–151.
Mead, C. (1989). Analog VLSI and neural systems Addison-Wesley.
Miyajima, T., Hasegawa, T., & Haneishi, M. (1993). On the multiuser detection using a neural network in code-division multiple-access communications. IEICE Trans. on Communications E76-B, 961–968.
Mostafa, M. (2014). Equalization and decoding: a continuous-time dynamical approach. Ph.D. thesis, Ulm University Institute of Communications Engineering.
Mostafa, M., Teich, W.G., & Lindner, J. (2012). Vector equalization based on continuous-time recurrent neural networks. In 6Th IEEE international conference on signal processing and communication systems (pp. 1–7). Australia: Gold Coast.
Mostafa, M., Teich, W.G., & Lindner, J. (2014). Approximation of activation functions for vector equalization based on recurrent neural networks. In 6Th international symposium on turbo codes and iterative information processing (pp. 52–56). Germany: Bremen.
Parlak, M., Matsuo, M., & Buckwalter, J.F. (2012). Analog signal processing for pulse compression radar in 90-nm CMOS. IEEE Transactions on Microwave Theory and Techniques, 60(12), 3810–3822.
Prometheus-GmbH: Top500 list (2015). http://www.top500.org [Accessed on December 2015].
Schlottmann, C.R., & Hasler, J. (2014). High-level modeling of analog computational elements for signal processing applications. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 22(9), 1945–1953.
Teich, W.G., Engelhart, A., Schlecker, W., Gessler, R., & Pfleiderer, H.J. (2000). Towards an efficient hardware implementation of recurrent neural network based multiuser detection. In IEEE 6Th international symposium on spread spectrum techniques and applications (pp. 662–665). New Jersey, USA: NJIT.
Teich, W.G., & Seidl, M. (1996). Code division multiple access communications: multiuser detection based on a recurrent neural network structure. IEEE 4th International Symposium on Spread Spectrum Techniques and Applications, 3, 979– 984.
Acknowledgments
Financial support by the German research foundation DFG (Deutsche Forschungsgemeinschaft) is gratefully acknowledged. A note of thanks goes also to IHP GmbH for the Si/SiGe foundry processes needed to realize the circuit.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Oliveri, G., Mostafa, M., Teich, W.G. et al. Advanced Low Power High Speed Nonlinear Signal Processing: An Analog VLSI Example.. J Sign Process Syst 89, 163–180 (2017). https://doi.org/10.1007/s11265-016-1171-0
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11265-016-1171-0