Let us discuss the various hardware components involved in pervasive computing devices.
The evolution of mobile devices suitable for pervasive computing is under progress. The technologies that support the evolution are considered in three categories, namely, as follows:
1.Hardware technologies.
2.Software technologies for human-machine interaction.
3.Operating systems and Java platforms for PDA Let us discuss these in detail.
Hardware
The size of input and output components {keyboard and LCD} decides the overall size of the mobile device. However, these cannot be smaller than the mobile phones that are available today. We may have smaller input devices that may be a part of the fabric worn by us or hidden in glasses or pens or jewellery. There are other hardware components requiring to be designed especially for mobile applications.
Batteries
Nickel- cadmium {NiCad} rechargeable batteries are being used in majority of mobile applications. These suffer from the disadvantage of being heavy and having a tendency to lose capacity due to memory effect Newer technologies such as Nickel-metal hydride {NiMH} and lithium ion {Li ion} are being accepted due to better capacity with less weight Lithium ion batteries are lighter with better energy density and so are used in majority of electronic equipment. The recent mobile phones are designed with reduced power requirements making the batteries to offer longer talk times.
Displays
LCDs are preferred because they are more readable and have benefits in terms of weight size and power consumption. The dual scan {DSTN} displays consume less power than the thin-film transistor {TFT} displays. Newer and more promising technologies are light-emitting organic diode {OLED} and light-emitting polymer {LEP} technologies. OLEDs can be built in almost any size and colour. Today they are available at commercially attractive prices. With these extremely thin displays can be made for any shape. Other newer technologies are chip-on-glass {CoG} and liquid crystal -on-glass {LCoG}. These can be used in the manufacture of very small display with a pixel size of 10 micrometers. These displays require some magnification.
Memory
Memory plays an important role in any computing device. The price of memory has seen downward trend over the years. One can see that at current technologies and costs, it is possible to have memory of several megabytes in mobile hand-held devices such as smart phones, digital cameras, MP3 players and PDAs. Due to restrictions of space (size) and power consumption, mobile devices cannot have hard disks as in PCs for storage. Though smaller removable hard disk drives are available now (with a capacity extending up to 1 GB), mobile devices tend to use in non-volatile Flash memory or battery backed up RAM because of power and access time requirements.
Typically the mobile devices can have memory up to 16 MB built-in. Expansion slots are available with which this can be enhanced. The expansion memory modules can be used to exchange the data in them to be transferred to CD ROMs or Hard disks in a PC. Static RAMs (SRAM) are preferred over dynamic RAMs (DRAM) as they do not require refresh cycles. Recently introduced Uni-transistor RAM (Ut RAM) is able to offer higher capacity memory chips in small size requiring less power. Other options are Magneto-resistive RAM (MRAM) and ferroelectric RAM (FRAM). These are similar to SRAM (requiring no refresh cycles) but consuming lesser power.
Processors
The processors have undergone sea change over the years. Clock rates and the number of transistors per chip have increased considerably. Operating voltages of the processors have come down from 3.3 V in 1995 to 1.35 V in 2000 resulting in lesser power consumption and lesser generation of heat. So, we have processors that are found in PDAs with small size and more power. Intel has introduced Speed step technology with which it is possible to switch on / off parts of the processors depending on the need fro them in present calculations. It is also possible to change the speed of the processor and core voltage depending on the available supply voltage. With this technology, the processor can switch between two modes of operation; one, when the processor is powered by power supply and the other when the processor is powered by the battery.
When powered by supply, the processor operates at full clock speed with full core voltage resulting in increased performance. When operating on battery the processor cuts down the clock speed and core voltage resulting in reduced performance. The processor can switch from one mode to the other depending on the power source and the transition will not be felt by the user. Another approach to power reduction is through reduction of transistors in the processors and replacing their functions with suitable software.
The software dynamically translates the original instructions in to modified instructions. Crusoe processor that is being offered by Transmeta is an example of such a processor. It has small core designed for 128 very long instruction word (VLIW). It can execute up to 4 operations per cycle. The morphing software emulates x86 compatible processor. During boot, the processor loads its software in to main memory. The benefit of using Crusoe processor is that it can be made to emulate any processor. It comes with an advantage of consuming only a few watts at the cost of high memory requirements for morphing software.
Intel's Speed Step technology
Recent processors include improvements in power management. These processors are capable of changing their internal clock frequencies and core voltage to adapt to changes in power supply.
Newer designs are even capable of switching parts of the CPU on or off depending on whether the current calculations require them to be available.
One such design is the Speed Step technology from Intel.
While the system is connected to an external power supply, the full clock rate and core voltage are available to the processor, resulting in the maximum performance.
When running on batteries, the clock rate and core voltage of the processor are reduced, resulting in significant power savings.
The transition between both modes is very fast and completely transparent to the user.
During the boot cycle, the Crusoe processor loads its software into a section of the main memory.
Frequently used code parts are optimized during run-time and kept in a separate cache.
A technology called LongRun promises to reduce the power consumption even more by reducing the processor's voltage on the fly when the processor is idle.
The big advantage of this approach is that the Crusoe processor can be used to emulate almost any other processor and uses only a few watts, even with high clock rates.
5.4 Human-machine Interfaces
Let us discuss the various issues and challenges for interfacing the human machine in pervasive computing.
Since the size of the mobile device is an important factor, the devices tend to have keyboards and displays in small form factors to suit particular device. The keys are kept to an absolute minimum using them only to trigger a menu or navigate a menu. PDAs and mobile phones are good examples of such devices. Some of the devices may not have any keyboard or display at all. These are ‘headless’ devices. These are used as either controllers or interfaced to other devices. Some of the functions of such devices are discussed.
Navigation
Many mobile devices have different cursor keys for navigation such as the one we have as an integrated cursor key for four-directional navigation in mobile phones. This key can be used to navigate up, down, right or left in a menu. One preference we have is the ability of operating the button with the thumb while keeping the device on hand / palm. Normally the navigation is by pressing any one of the four sides and the selection is through pressing the center. These navigation buttons offer haptic (the sense of touch) feedback to the user. The Navi Roller navigator in Nokia 7110 is shown in Figure. Similarly Sony has introduced Jog dial (in CMD CD 5 phones) for simple navigation for their built-in applications.
Navi Roller on a Nokia 7110 model
Haptic Interfaces
A single force-feedback knob can emulate the feel of conventional control knobs (detents, limit stops, friction) and can produce new effects as well such as vibration, scrolling, and free-spin, all instantaneously reconfigurable under computer control. Such a haptic knob can simulate the functions of all instrument controls that it replaces and thus reduced to one device, the control of the dashboard provides the driver with instant access to all functions, quickly and ergonomically. A haptic knob that additionally incorporates a brake to provide high torque capability in a small volume with low power consumption is demonstrated. The device consists of a motor with optical encoder, a brake, a flexible coupling, a micro switch and an instrument knob. Figure shows the picture of the experimental prototype that was built to demonstrate the same.
The haptic knob device communicates with a computer through A/D and encoder boards. The haptic knob simulates the functions of the knobs and push buttons from the dashboard of a vehicle. Some of the functions programmed are brake detents, seek, balance-brake, Fan speed and constant speed.
Rotary Haptic Knobs for Vehicular Instrument Controls
Similarly, a programmable rotating actuator with haptic feedback is designed and implemented by VDO. It acts both as rotating control device and a push button. There are sensors which detect the position of the knob and a motor produces equivalent torque. So, the motor rotates when the knob is rotated. This allows a single knob to be used for selecting many options from a menu.
When the same type of navigation control is used in a car, the driver can have a ‘feel’ of the actual position and the driver can change it without having to have a look at the control. The set up has programmable option that allows it to be used to control many settings with different number of options to choose from.
Keyboards
We have seen that the preferred size of the mobile device dictates the size of the keyboard. Depending on the size of the keyboard allowed within the mobile phone form factor, there may be full set of keys or a limited set of keys in the keyboard. If the standard ‘qwerty’ type of keyboard is preferred, the size of the device becomes larger. Making the user to use only numerical pad to enter alphanumeric characters will make the entry cumbersome.
In some cases, the form factor or the usage environment may not allow any keyboard to be present in the device. In such cases, the device may have other input technologies such as handwriting / voice recognition software. AS an option some mobile phones use on-screen keyboards. Figure shows the Motorola mobile phone model Crush that uses touch screen.
Motorola Crush – Touch screen phone
It can be seen that the phone does not have any keyboard. Instead, it uses a touch screen or entering the text / number. Figure shows the on-screen keyboard of a Palm PDA.
The Palm on-screen keyboard
Numbers and special characters are entered using another screen in another mode. Another keyboard type is the ‘fitaly’ keyboard. This is different from the ‘qwerty’ keyboard that is used in typewriters. The arrangement of the letters in ‘fitaly’ keyboard is based on the frequency of usage of each letter. Most frequently used letters are placed in the middle columns and rows. Figure shows the ‘fitaly’ keyboard layout. The name is derived from the letters on one of the top rows. This keyboard can be displayed on screen in place of ‘qwerty’ keyboard. This keyboard is found to be useful for using a single finger for typing. This can support up to 220 characters of ANSI / ISO Latin-character set.
‘Fitaly’ keyboard layout
Numeric Pad Keyboard is other type of Keyboard. This type is used in many mobile devices. This makes best use of keyboard space. There are numeric keys and these numeric keys are used for entry of text also. Each number key is associated with a few letters of the alphabet. For example, number 3 is associated with letters d, e and f. If this key is pressed once, d will be entered. If it is pressed twice, e will be recognized. When pressed 3 times, f is entered. If it is pressed 4 times, ‘3’ is entered.
The Figure shows the layout of Tegic T9 input system. In this input system each key is pressed only once. The software identifies the world associated with the set of entered keys. Sometimes, same type of key strokes may throw up choice of more than one word such as SNOW and PONY. At such a time, the user is given an option to choose from the available options.
Tegic T9 input system in Plam PDA
Handwriting recognition
Handwriting recognition is feasible today with the available touch sensitive displays and processing power of the processors used. The ability to recognize handwriting depends mostly on the way the letters are written (cursive or individual lexers printed), available processing power and the precision of the input mechanism. The most difficult and expensive method is recognizing the completely handwritten word. While it is most natural and easy way for the user to enter, it is the most difficult to correctly identify the word written:
CalliGrapher on a Psion Series 5
Figure shows the entered text and recognized word with the Calligrapher software on Psion series 5 device. The draw backs of this approach are that it is delayed in response and requires precise data capture. There are other methods in which each character is recognized. In this case each letter must be written separately leaving the stylus not touching the writing surface between characters. Though the rate of recognition in this technology is better, it requires specific efforts by the user by way of lifting the stylus after every letter.
Graffitti help on the Palm
Figure shows the Graffiti input method in Palm PDA In this case, the user has to follow specific writing style as specified. Some methods allow the users to train the software to understand individual’s handwriting. In this case, if a person other than the person who trained the system writes, fresh training must be undertaken and a separate new profile must be created. Some of the Asian languages are based on a set of symbols. These cannot have input methods discussed above. These languages have tens of thousands of symbols. There are separate input methods existing for entering these symbols. These methods are based on combination separate strokes whose combination forms the required phonic form.
Input methods available with CJKOS
Figure shows CJKOS that is an OS extension on Palm OS especially for Chinese, Japanese and Korean languages.
Speech Recognition
Speech recognition is the most advantageous and expensive natural input method into mobile devices. Many computers have the capability of speech recognition. It is expected that these capabilities will be made available very shortly in mobile devices also. The research in this area is in advance stage. Some mobile phones allow the user to select an entry from the address book just by voice by speaking the name. Further advancement such as complete operation by voice, complex queries and translation are expected shortly. Speech recognition will also help the driver to use the phone safely while driving. In its basic form speech recognition aims recognizing numerals and a very small set of words. However continuous speech recognition is a more complex process. More processing power is required for the same to be available on a mobile device. Specialized hardware and processors are likely to be available.
5.5 Biometrics
Let us see how biometric signals are used in pervasive computing.
Personal characteristics such a figure print, iris scan, signature, hand geometry, face recognition and voice recognition are used fir biometric authentication. Since they use small sensors, fingerpoint, speaker verification are used in pervasive computing.
Biometric authentication
Biometric authentication system capture the user’s characteristics with a sensor, derive characteristic values, and compare this with a known reference. The result of the comparison is either 0, if the authentication was not successfully performed, or 1, if authentication was successfully performed. The image system gets the bifurcation points of finger line and compare with the existing one stored in the system. Concept is shown in the Fig.