Carl Zetie and Ken Smiley
Fundamental limitations in current PDA technology rule out some things we’d
like to do. How fast can we expect things to improve?
The technology that makes up a complete mobile or wireless solution consists
of a complex stack of hardware, software and infrastructure components, each of
which is on its own cycle for improvements. Because personal digital assistants
(PDAs) are often pushing the limits of size, weight and power, these limits are
often more significant to PDA deployments than they are to typical desktop or
laptop-based applications. Some of the key cycle times are as follows:
- Minor software upgrades: Every six months to one year. These provide
incremental improvements in capability, perhaps adding support for new
markup languages or new hardware interfaces such as adding SD support. Like
PCs, PDAs may increasingly get virus definition upgrades on a near-daily
basis.
- Major software/operating system upgrade: Every two to four years. These
provide significant improvements. For example, Pocket PC 2002 added greatly
improved security and mandatory upgradeable flash memory over previous
versions, and PalmOS 5.0 will provide greatly enhanced multitasking. Upgrade
cycles have accelerated recently and may come as fast as every 18 months in
the future. Note, however, that operating system upgrades are increasingly
closely tied to hardware upgrades as the two are become increasingly
intertwined.
- Memory (RAM) size: Doubling every year. Fundamental improvements in memory
technology are on the same 18-month to two-year cycle as for PCs, but the
memory available in PDAs is also tied to the cost of memory, which has a
cycle all its own that goes up as well as down. On average, the
"typical" memory in PDAs has been more than doubling every year,
with the maximum available memory growing slightly more slowly.
- Processor speed: Incremental increases every three to six months.
Discontinuous increases every two to four years. Two different cycles affect
processor speed. While a PDA stays within a single processor family,
improvements come in comparatively small increments, and the rate of
increase depends on whether faster processors are available for immediate
adoption and whether faster versions can be added. For example, during the
past three to four years, Palm PDAs have moved from 16Mhz Motorola
Dragonball 68K CPUs through the 33Mhz processors found in most current
devices to the 66Mhz processors found in the fastest devices. Meanwhile,
Pocket PC-based devices have standardized on the StrongARM processor running
at 206Mhz and no longer support MIPS or SH3 processors. More dramatic
increases can occur when a PDA moves from one technology to another, for
example Palm’s move from Dragonball 68K to ARM in the forthcoming PalmOS
5. Such discontinuous increases are typically tied to the major operating
system upgrade cycle. Note, however, that some devices may choose not to
exploit the full speed of a given processor in order to reduce power
requirements and improve battery life. Also, without supporting changes in
the operating system, a new processor alone may not provide a performance
boost, as has befallen the new 400Mhz XScale processors in the latest
generation Pocket PC devices. Consequently, these two cycles need to be
considered jointly.
- Mass storage: In the past, the rate of increase of technologies such as
SmartMedia or Compact Flash Memory has been roughly to double every year.
With Microdrives from vendors such as IBM now offering 5GB or more of data
in a Compact Flash format, and solid state Compact Flash memory offering
1GB, availability of mass storage should very rarely be a limiting
technology today. Toshiba has a 10GB drive in PCMCIA format, expected to go
to 20GB next year. SD memory continues to trail behind other formats in
size, currently topping out at 128MB; however, it should make up the
difference following a slightly accelerated schedule as compared to other
formats.
- Network bandwidth: Ten-fold increase every three to five years. Bandwidth
on the cellular networks has almost the slowest and the most discontinuous
change cycle. Data bandwidth is tied to the generational network upgrades of
the wireless carriers, and these are massive infrastructural changes that
take place over three- to five-year periods. Under 2G digital mobile
networks, 14.4Kbps was a normal maximum for mobile data, perhaps achieving
28.8kbps equivalent with software compression. Under the 2.5G networks
currently being rolled out (CDMA 1xRTT and GSM/GPRS), expect 25Kbps to
70Kbps average performance without compression, increasing gradually to
80Kbps, potentially peaking at 150Kbps. Future 3G networks promise the
possibility of up to 2Mbps when stationary, less when moving, but will not
be widely available until 2005. Some intermediate relief may be offered by
the growth of "hot spots" along with seamless LAN-to-WAN roaming,
but such technologies are geographically very limited (see Planning
Assumption, Wireless LAN to WAN: Solutions Beginning to Appear, Stan Schatt).
- Battery life: Fundamental power technology has changed very slowly
compared to these other technologies, although significant incremental
improvements in size, weight and life have been driven in recent years by
demands of the cell phone industry, among others. The arrival of dramatic
improvements in power technology is unpredictable, but incremental
improvements in battery life in the near term will come more predictably
from better components, including more efficient displays, CPUs with better
power management as epitomized by the Transmeta Crusoe and Intel XScale, and
smarter software that can manage the power consumption of the system
according to need. Supplemental power sources are becoming increasingly
available as the number of mobile devices in use drives demand, including
user-swappable batteries, power sockets on airplanes, hand-wound and solar
chargers, etc. Giga’s Power 2002 conference will discuss these topics in
detail.
By identifying the fundamental limitations that make a particular deployment
infeasible today, and understanding the cycle time associated with that
technology, you can better understand the trade-off involved between
compromising with today’s technology and waiting for something better to
arrive. Given these independent cycles, Giga advises re-evaluating the
suitability of the PDA as a platform at least once a year if it has been deemed
to be inadequate in the past.
(Contributing Analysts: Rob Enderle and Brownlee Thomas.)