Advances in Neural Interfacing
Research in systems that interface with the central and the peripheral nervous system (CNS and PNS respectively), can be categorized in three main areas: neural recording, biosignal processing and neurostimulation. Devices for monitoring neural activity have a range of applications including neurophysiology and neuropathology studies, diagnostics (employing biosignal processing algorithms) and they can be used for rehabilitation (as feedback to neurostimulation) [1]. Functional electrical stimulation (FES) has been developed in various degrees for treatment of conditions including foot drop, hand grasp, spinal cord injury, cochlear and retinal diseases, pain relief and epilepsy, to name a few [2-4]. 
Today, there is a massive surge in neural interfacing research, mainly due to recent advances in microelectronic and electrode technology. Various platforms been developed both for recording and stimulation, ranging from external surface-electrode devices to implants. Companies and academic institutions seek formulas of combining expertise from different disciplines, including microelectronics, biology, medicine and chemistry forming a trend for interdisciplinary research. 
This article gives an overview of some of the main areas and current developments of non-invasive PNS interfacing and CNS biosignal processing, with a focus on novel methodologies for advancing the field.
Selective and ionic peripheral nerve recording: Cuff electrodes have been widely used for nerve signal (electroneurogram or ENG) recordings because they are biocompatible and suitable for chronic implantation as they are non-invasive to the nerve [5,6]. However they are greatly limited by the inability to identify specific nerve fibre groups (fascicles) that are active inside a bundle during a measurement. Fascicle selectivity would allow the association of the recorded neural signal from centrally-implanted cuffs with targeted muscle-groups or organs. Current studies on selective recording employ “split-electrode” or “multi-electrode” configurations and have reported various degrees of selectivity [7,8]. However, these studies have not considered the severity of electrode noise, which is a very serious limitation to small-area electrodes, as they employed stimulation-induced nerve signals. Such signals are usually much larger than microvolt-scale naturally occurring ENG. Thus, further research is required to achieve satisfactory levels of fascicle selectivity in recordings of naturally-occurring ENG. Other desirable selectivity forms include sensory-motor signal discrimination and fibre diameter classification.

Fig. 1: An artist’s impression of structures developed by the Advanced Neural Interfaces group at Institute of Biomedical Engineering, Imperial College London. (a) Multiple sensor array on flexible substrate, forming a chemFET cuff [9] (b) An array of ion-specific ISFETs on monolithic silicon substrate.

As nerve conduction is based on ionic current flow, potassium and sodium sensing can be used as part of a novel neural recording platform offering selectivity with satisfactory SNR, myoelectric interference rejection, and overall miniaturization of the interface [9]. Most neural monitoring research so far has focused on the electrical aspect of the signal with very few studies mentioning ionic sensing of this sort [10]. This is almost entirely due to the conventional apparatus for ion sensing in chemical laboratories being cumbersome and inappropriate for biosignal recording [11].

Fig. 2: Neuroprostetic ICs developed by the Advanced Neural Interfaces group at Institute of Biomedical Engineering, Imperial College London. (a) Reconfigurable multi-channel stimulator encompassing a neurochemical sensor array, an asynchronous artificial-neural bus and an interleaved biphasic stimulus generator. (b) Single potassium-sensitive ISFET “pixel” with local processing circuitry.

ISFET (ion-sensitive field effect transistor) - based chemical sensors, offer an attractive solution for ion sensing of this kind due to their minimal size, sensitivity, fast response and prospect of monolithic integration [12]. Figure 1a depicts an implementation of ion-specific ISFETs (or chemFETs) as part of a recording chemFET cuff, as proposed in [9]. In a solution presented in Figs. 1b and 2a, custom-designed ISFETs based on CMOS technology have been integrated with compact local processing including bias, drive and signal conversion circuitry to realise “intelligent chemical sensors” (Fig. 2b) for neuro-chemical detection from nerve bundles. Their surface was coated with appropriate ionophores to make them potassium-sensitive [13]. A multitude of such sensor “pixels” was placed in an array, providing spatio-temporal selectivity and sensor failure minimization through redundancy. The ADC and digital post-processing stages often used in neural amplifier schemes, have been replaced by low power “integrate-and-fire” neuromimetic circuit encoding the sensor information in the spike domain with multiple sensor fusion and off-chip transmission facilitated using a standard AER (Address Event Representation) protocol.
Selective peripheral nerve stimulation: Desired features of advanced neurostimulation schemes of a nerve bundle include fibre-diameter selectivity, unidirectional stimulation (motor or sensory), and fascicle selectivity. Research on implementing such features in FES devices has been extensively carried out in spinal cord injury (SCI) studies. However, research in other areas, such as epilepsy treatment, has not as yet benefited from expertise developed for SCI stimulators because advances in neuroprostheses have not developed coherently [14]. Although well-established for chronic implantation, PNS interfacing for epilepsy treatment lacks selectivity, thus reducing therapeutic effectiveness and causing considerable side-effects to the patient. It is thus desirable to improve epilepsy platforms through selective stimulation, by developing a novel approach to selectivity. Such novelty can be achieved by exploiting the natural physiological properties of the vagus nerve and the nodose ganglion in order to harness the body’s innate ability for directional communication.  The therapeutic power and accuracy afforded by such a design has potential to improve vagal platforms for seizure-specific epilepsy treatment.
Algorithms for EEG diagnostics: It is well understood that one of the most challenging aspects of Epilepsy therapy is that ability to understand and predict the onsets of seizures. Since the 1970's a vast amount of work has been developed on the capability of the electroencephalogram (EEG) in reaching these therapeutic goals. 
The research has gone from simply identifying dynamics of the signal that could act as precursors to the onset of a seizure to fully tested algorithmic implementations based on a number of data testing mechanisms (e.g. surrogate data testing). There is still a large amount of work to be done [15,16]. A number of current developments have led to more robust and unified testing mechanisms which will allow better comparative analysis of the results obtained from the currently available algorithms. Based on this and a number of other technological, mathematical and experimental developments in epilepsy research, algorithms with the potential for prediction of seizures should be identified with the future goal of hardware implementation as part of a neural platform. In correlation with PNS interfaces, this platform will offer a new insight into the treatment and understanding of epilepsy.
[1] J. J. Struijk, M. Thomsen, J. O. Larsen, and T. Sinkjaer, "Cuff electrodes for long-term recording of natural sensory information," IEEE Eng. Med. Biol., vol. 18, pp. 91-98, 1999.
[2] M. Hansen, M. Haugland, T. Sinkjaer, and N. Donaldson, "Real time foot drop correction using machine learning and natural sensors," Neuromod., vol. 5, pp. 41-53, 2002.
[3] A. Inmann and M. Haugland, "Implementation of natural sensory feedback in a portable control system for a hand grasp neuroprosthesis," Medical Eng. & Phys., vol. 26, pp. 449-458, 2004.
[4] M. S. Evans, S. Verma-Ahuja, D. K. Naritoku, and J. A. Espinosa, "Intraoperative human vagus nerve compound action potentials," Acta Neurol. Scand., vol. 110, pp. 232-238, 2004.
[5] L. N. S. Andreasen, J. J. Struijk, and S. Lawrence, "Measurement of the performance of nerve cuff electrodes for recording," Medical & Biological Engineering & Computing, vol. 38, pp. 447-453, 2000.
[6] I. F. Triantis, A. Demosthenous, and N. Donaldson, "On cuff imbalance and tripolar ENG amplifier configurations," IEEE Transactions on Biomedical Engineering, vol. 52, pp. 314 - 320, 2005.
[7] J. J. Struijk, M. K. Haugland, and M. Thomsen, "Fascicle selective eecording with a nerve cuff electrode," 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Amsterdam, 1996.
[8] B. K. Lichtenberg and C. J. De Luca, "Distinguishability of functionally distinct evoked neuroelectric signals on the surface of a nerve," IEEE Trans. Biomed. Eng., vol. 26, pp. 228-237, 1979.
[9] I. F. Triantis and C. Toumazou, "Advanced mixed-mode nerve cuff interface," UK patent, patent no. GB 0613698.0 - patent pending, 2006.
[10] R. M. Siegel and R. I. Birks, "A slow potassium conductance after action potential bursts in rabbit vagal C fibers," Am. J. Physiol. Regulatory Integrative Comp. Physiol., vol. 254, pp. 443-452, 1988.
[11] P. Bergvelt, "Thirty years of ISFETOLOGY: What happened in the past 30 years and what may happen in the next 30 years," Sensors and Actuators, vol. 88, pp. 1-20, 2003.
[12] P. A. Hammond, D. Ali, and D. R. S. Cumming, "Design of a single-chip pH sensor using a conventional 0.6-mu m cmos process," IEEE Sensors, vol. 4, pp. 706–712, 2004.
[13] A. Bratov, N. Abramova, C. Dominguez, and A. Baldi, "Ion-selective field effect transistor (ISFET)-based calcium ion sensor with photocured polyurethane membrane suitable for ionised calcium determination in milk," Analytica Chimica Acta, vol. 408, pp. 57–64, 2000.
[14] P. R. Troyk and N. de N. Donaldson, "Implantable FES stimulation systems: what is needed?" Neuromodulation, vol. 4, pp. 196–204, 2001.
[15] C. J. Stam, "Nonlinear dynamical analysis of EEG and MEG: Review of an emerging field," Clinical Neurophysiology, vol. 116, pp. 2266-2301, 2005.
[16] F. Mormann, R. Andrzejak, C. E. Elger, and K. Lehnertz, "Seizure prediction: the long and winding road," Brain, vol. 130, pp. 314 -33, 2007.
Iasonas F. Triantis*, Virginia Woods*, Amir Eftekhar*, Pantelis Georgiou*, Timothy Constandinou*, Emmanuel M. Drakakis** and Chris Toumazou*, *Advanced Neural Interfacing Group, Bionics Team, Institute of Biomedical Engineering, **Department of Bioengineering, Imperial College London, UK (Email: i.triantis@imperial.ac.uk)

On-Chip Inductor Synthesis and Design for RF Applications: ASITIC vs. FastHenry
A typical RFIC receiver front-end comprises an LNA, mixer, frequency synthesizer, channel filtering and baseband/IF amplifiers. Depending on the required noise figure (NF) for the wireless application, the LNA may or may not be integrated with the rest of the subsystem. With the proliferation of commodity wireless standards from GSM to WCDMA for mobile cellular communication, BlueTooth to ZigBee for wireless personal area networks (PANs), WiFi to WiMAX for wireless local and metropolitan area networks (LANs and MANs), and DVB-H, DAB, ISDB-T and T-DMB for mobile video and audio broadcast, LNA integration is a necessity from a system cost reduction perspective. These LNAs typically require on-chip inductors for input impedance matching and to provide a resonant load circuit in order to deliver sufficient gain at the desired RF.  These inductors are typically designed using high-end commercial electromagnetic simulators. An alternative is the use of freeware electromagnetic simulators of which two popularly employed examples are ASITIC [1] and FastHenry [2]. But are these freeware simulators sufficiently accurate? The rest of the article focuses on integrated inductor impairments at RF. Subsequently, an inductor is designed for use in a low noise amplifier using both ASITIC and FastHenry. Finally, the measured LNA gain is compared to simulated LNA gain using both inductor models with ASITIC results most closely matching measurements.
Inductor quality factor at RF: An on-chip inductor is typically implemented using the top two metal layers of the process technology with tungsten vias used to connect metal layers together. A conductive shield is generally used below the inductor to terminate the electric fields and significantly reduce resistive substrate losses. Fig. 1 shows the inductor outline with the model used for inductor simulation at the right side of the figure. A square spiral outline is commonly used with the optimum outline being a cyclic spiral (this is however difficult to draw in an integrated circuit layout tool and is usually approximated using an octagonal spiral as in Fig. 1). The dc resistance of an on-chip inductor arises due to the resistivity of the inductor metallization and metal-to-metal vias. Being a device exhibiting reactance, the inductor impedance at frequency f is given by Zind = Rdc + Rac + j 2πf (Ldc + Lac), where Rac is the additional resistance at RF due to skin effect (electrons flowing only at conductor surface), proximity or current crowding effect (negative magnetic interactions between inductor turns close to each other) and eddy current effect (induced diametric substrate or shield current creating negative magnetic interactions with inductor that increases Rac). These effects also affect the dc inductance (Ldc) with Lac being an additional negative inductance reducing overall inductance at RF. The quality factor (Qind) of an inductor is the ratio of the electrical energy stored in the inductor to the dissipated energy and is given by Qind = 2πf (Ldc + Lac)/(Rdc + Rac). In order to simultaneously terminate the inductor’s electric field and reduce the swirling eddy currents in the substrate or shield, the shield is usually patterned [3] with a suitably patterned shield yielding an inductor with higher Q than the un-shielded inductor at the self resonance frequency.
FastHenry vs. ASITIC: A five-turn octagonal inductor was synthesized using FastHenry 3.0 for Windows [4] and ASITIC for Cygwin on Windows [5]. A 3.2GHz AMD Athlon PC running Windows XP was used with simulation time being a few seconds for both application when the recommended FastHenry discretization filament sizes [4] and ASITIC fft sizes [5] are used. A patterned ground plane was also defined in the substrate layer for the inductor. The inductor was made from the top two metal layers in the technology. Target inductance was about 4.15nH at 2GHz.
Results: Both tools yielded inductance close to the target value at 2GHz (FastHenry 2GHz L: 4.19nH, ASITIC 2GHz L: 4.22nH). Respective dc resistance and inductance are 3.8 & 4.23nH for FastHenry and 3.6 & 4.31nH for ASITIC. ASITIC yielded higher overall 2GHz resistance as it better accounted for eddy current effects (FastHenry 2GHz R: 4.6, ASITIC 2GHz R: 7.6). Corresponding Q values for FastHenry and ASITIC are 11.44 and 6.97 respectively. Fig. 2 are plots comparing measured gain response of the LNA with simulated gain response using inductors modeled with FastHenry and with ASITIC with the ASITIC simulated results being closest to measurement.
In conclusion, both FastHenry and ASITIC yield inductance values to within ± 2% with FastHenry slightly more accurate (about 1%). However, ASITIC seems to better model additional ac resistance due to eddy currents with ASITIC-simulated results closely matching laboratory measurements.
[1] M. Kamon, M. J. Tsulk, and J. K. White, “FASTHENRY: a multipole accelerated 3-D inductance extraction program,” IEEE Trans. Microwave Theory Tech., vol. 42, no. 9, pp. 1750-1757, Sept. 1994.
[2] A. M. Niknejad and R. G. Meyer, “Analysis, design, and optimization of spiral inductors and transformers for Si RF IC’s,” IEEE J. Solid-State Circuits, vol. 33, no. 10, pp. 1470–1481, Oct. 1998.
[3] C. P. Yue and S. S. Wong, “On-chip spiral inductors with patterned ground shields for Si-based RF IC’s,” IEEE J. Solid-State Circuits, vol. 33, no. 5, pp. 743–752, May 1998.
[4] http://www.fastfieldsolvers.com
[5] http://rfic.eecs.berkeley.edu/~niknejad/asitic.html
Olujide A. Adeniran and Anthony M. Eaton, Mirics Semiconductor Inc., UK (Email: jide.adeniran@mirics.com)

In Memory of Prof. Richard Newton
UC Berkeley News, Jan. 2, 2007 – “A. Richard Newton, professor and dean of the College of Engineering at the University of California, Berkeley, a pioneer in integrated circuit design and electronic systems architecture, and a visionary leader in the technology industry, has died. He was 55. Newton died today at the UC San Francisco Medical Center of pancreatic cancer.” 
The news from Berkeley was heard first with disbelief, then sadness, and finally even anger. How could someone so vital, so young, be taken from us? This Aussie with the infectuous grin and incredible energy who had added California, then America, and finally the World to his domain had so much more to accomplish! 
As a PhD student and then as a much-honored member of the UC Berkeley EECS faculty he and his students and colleagues developed many of the critical EDA tools that enabled the Integrated Circuit revolution – he organized the distribution of SPICE and other UCB tools, conceived EDA frameworks and managed the UCB industrial liaison program that developed the model for the close industry/academia relationship which is the hallmark of the IC revolution. Not content to limiting himself to academia, he was instrumental in founding and growing the companies who placed these tools in the hands of engineers worldwide. – Cadence, Synopsys, PiE, Simplex among them. In 2003, Newton won the EDA industry's highest honor, the EDA Consortium Phil Kaufman award. Asked what he viewed as his greatest contribution to EDA, Newton answered "it's my students. I'm proud of all my students and what they've gone on to do — it's a bit of a who's who in EDA."  (EE Times, 01/02/2007 Richard Goering)
In 2000, UC Berkeley promoted Rich to be Dean of the Engineering. He now had the means to implement his vision, becoming a passionate advocate for the use of technology to tackle some of society's most difficult challenges, particularly those in developing nations.  He was the driving force behind a new organization at Berkeley called CITRIS - Center for Information Technology Research in the Interest of Society.  CITRIS has now expanded to 3 other University of California campuses.
Newton became a champion of synthetic biology, seeing the emerging field as the application of engineering principles to the life sciences. He played a major role in the establishment of the Berkeley Center for Synthetic Biology, as well as of the Synthetic Biology Engineering Research Center, or SynBERC, launched last year with a $16 million grant from the National Science Foundation.
Newton was a strong advocate of promoting women in engineering, and while he was dean, the number of women on the faculty at the College of Engineering nearly doubled from 15 in 2000 to 27 today. 
Tributes and remembrances flowed in, as the news spread. Among them, 
"Rich Newton was a man of incomparable vision," said UC Berkeley Chancellor Robert Birgeneau. "Dynamic and entrepreneurial, he understood the power of engineering and technology in entirely new ways, and he connected it to addressing society's toughest problems.”
Orville Schell, UC Berkeley dean of the Graduate School of Journalism and a close family friend of Newton's, said Newton "was one of those most rare of men who was as kind and collegial as he was intelligent, energetic and competitive. His death leaves an emptiness of indescribable proportions."
"He had an unmatched capability of marrying technical insights with industrial needs," said Sangiovanni-Vincentelli. "He articulated the EDA roadmap 30 years ago, and almost all he said actually happened."
"Newton had an astute business mind, something you wouldn't necessarily expect from an academic," said Dado Banatao, managing partner of Tallwood Venture Capital and chair of the UCB College of Engineering advisory board. "There are a lot of visionaries out there, but when you have a visionary technologist, you understand how technologies can be applied to solve the right problems. Newton was a visionary technologist."
“An imposing figure at somewhere over six feet, Rich was literally “larger than life,” while his physical stature was still dwarfed by the height of his passion.   Rich did not wrestle with ideas; he conquered them.  He entered any room, any conversation, or any brainstorm session with full enthusiasm and a wonderful sparkle of irreverence.  Be it a new EDA algorithm, a communication system for rural India, a book on innovation, or a promising bottle of red wine, all got the full assault of his “joie de vivre.” Aart de Geus, Chairman & CEO, Synopsys.
We’ll all miss Rich immensely.
Chuck Shaw, Cadence Design Systems, USA (Email: shaw@cadence.com)
                                                                                                                                                                                Technology & Research News_files/Fig1-1.jpgTechnology & Research News_files/Fig2-1.jpgmailto:i.triantis@imperial.ac.ukTechnology & Research News_files/Article_Fig1-1.pngTechnology & Research News_files/Article_Fig2-1.pnghttp://www.fastfieldsolvers.comhttp://rfic.eecs.berkeley.edu/~niknejad/asitic.htmlmailto:jide.adeniran@mirics.com?subject=mailto:shaw@cadence.comshapeimage_3_link_2shapeimage_3_link_5shapeimage_3_link_6shapeimage_3_link_7shapeimage_3_link_8
Fig. 1
Fig. 2