Human-Embedded Computing


Introduction to human-embedded systems

Nowadays a variety of embedded systems is used for medical and healthcare purposes. They range from wearable monitoring devices to brain implants.

Embedded systems in general are computer systems that are different from other types of computers such as PCs because they:


Human embedded systems have to match some additional requirements:


Implantable devices can be divided in two groups. The first group consists of stimulatory devices, such as pacemakers, implantable cardioverter/defibrillator, neurostimulators and pain suppression devices, Cochlear implants/hearing aids, etc. The second group consists of measurement/control devices, such as drug infusion and dispensing devices, artificial heart and heart assist devices, implanted sensors, etc.


Cochlear implant (bionic ear) as an example of an implantable embedded device


Deafness is most often caused by degeneration or loss of hair cells in the inner ear, rather than a problem with the associated neurons. This means that if the neurons can be stimulated by a means other than hair cells, some hearing can be restored. A Cochlear implant directly stimulates auditory nerves inside the cochlea, the snail‑like organ in the inner ear, to provide information about sound to the brain [4]. The implant was developed in 1978 by Graeme Clark and it was referred to as the bionic ear. The Cochlear implant consists of external and internal parts.

The external parts include a microphone, a speech processor, and a transmitter. The microphone picks up sounds and sends them to the speech processor. The speech processor is a computer that gets input from a microphone, analyzes and digitizes the sound signals according to the desired speech processing strategy, unique for every user. The output of the speech processor goes to a transmitter worn on the head just behind the ear.

The internal (implanted) parts include a receiver surgically implanted on the scull just under the skin and an array of electrodes surgically inserted in the cochlea. The receiver takes the coded electrical signals from the transmitter and delivers them to electrodes. The receiver includes a mini computer that translates the processed sound information and controls the electrical current sent to the electrodes. The electrodes stimulate the fibers of the auditory nerve in cochlea. The implantable system is often referred to as Implantable Cochlea System (ICS) and it does not contain a power source. Instead, it receives its operating power from signals transmitted to it [5]. A short video explaining how Cochlear implant works can be found here.

Figure 1. Schematic diagram of a typical Cochlear system architecture [5]


Speech processor


The Cochlear speech processor must be programmed individually for each user. The programming is performed by an audiologist trained to work with cochlear implants. The audiologist sets the minimum and maximum current level outputs for each electrode in the array based on the user's reports of loudness. The audiologist also selects the appropriate speech processing strategy and program parameters for the user. The desired temporal sequences of output currents are stored digitally within the speech processor as a table of weight coefficients. The table has time and electrode pair rows and columns. The intersection of a given row and columns gives the magnitude, polarity, and time during which current is applied to a given electrode pair. At the beginning of each new stimulation cycle speech processor multiplies the magnitude derived from the acoustic signal during previous cycle by the weight factors from the table. The system employs two strategies. One is known as a “continuous interleaved sampler” (CIS) where only one electrode pair is stimulated at any instant of time. The other one is known as a “continuous analog” (CA) when more than one electrode pair is stimulated at any instant of time. An example of a simple CIS signal generated from that table is given in Figure 2.

Figure 2. An example of a simple Continuous Interlined Sampler (CIS) [5]


Data frame used with the system

A data frame used in a typical speech processor [5] is made up of 83 bits (Figure 3). It contains nine words with nine bits each, a parity bit and an end-of-frame bit. The clock rate is such that it takes 77 microseconds to process the data frame. The first eight words are data words and the ninth one is a control word. The control word can be used to set various functions within the ICS, e.g. electrode configuration. Each nine-bit data word consists of first eight amplitude bits and a polarity bit. Each word maps to the particular electrode pair. First word is associated with the electrode pair in contact with the basal end of cochlea, which receives high frequency information; the last word is associated with the electrode in the apex of the cochlea, which receives low frequency information.

Figure 3. Data frame format


Some of the newer Cochlear systems, such as Advanced Bionics developed the Harmony™ HiResolution® Bionic Ear System, enable people to also hear music by providing wider frequency range from 150 Hz to 8000 Hz [6].

A typical example of an RF transceiver used in modern Cochlear implants is Zarlink ZL 70250 radio chip [7], which has small dimensions and very low power consumption (Table 1).






Harmony sound processor

54 mm x13 mm x 21 mm


ZL 70250

2 mm x 3 mm

<2mA (active),

<500nA (idle)



The next generation of human-embedded systems is going to take advantage of human energy harvesting (body movements, body temperature) and environmental energy harvesting (RF waves, vibrations) [8]. This is especially important for implantable systems, where battery changes are not practical. Some of the components, such as the ZL70250 radio frequency (RF) transceiver, are already very low power, which makes them suitable for embedded systems that apply human energy harvesting.



Cochlear implant has all the characteristics of a human embedded system. The speech processor is made specifically to perform speech processing and must be customized for a specific user. Therefore, it has to be programmable. It needs to be unobtrusive, and some of its parts are implanted. On the other hand, it has most of the characteristics of a typical computing system. 




[1]   Embedded systems, Wikipedia,

[2]   Nicolas Mokhoff, Implants to Remake Medicine, EE Times, 12/18/06,

[3]   PR Newswire Europe Ltd.,”Zarlink launches high performance surge protection chips for pacemakers, implanted defibrillators, neurostimulators,”

[4]   Cochlear implant, Wikipedia,

[5]   Loeb et al., Multichannel Cochlear prosthesis with flexible control of stimulus waveforms, US Patent,

[6]   Harmony™ Sound Processor Technical Specifications, Advanced Bionics,

[7]   ZL70250, Ultra Low Power RF Transceiver, Zarlink,

[8]   CITRIS Project: Energy Harvesting for Biomedical Devices and Health Care Intelligent Infrastructure, UC Berkeley, UC Davis,


Useful links

·         ULP Audio Transceiver (ZL70262), Zarlink,

·         Matlab Signal Processing Blockset, Cochlear Implant Speech Processor, The MathWorks,

·         Zarlink Radio Solution Powers Miniaturization of Medical Wireless and Remote Sensor Devices

·         Energy harvesting,

·         How a Cochlear implant works, Harmony Bionic Ear, September 20, 2008