Untitled

24 Sanjay Sarma
of a temperature sequence. In other words, the string of temperatures could be
compressed to a single representative number that captures the kinetics of the
chemistry. 2 An alternative is to use traditional compression. The third problem is one
of uploading the temperature log. RFID has limited bandwidth, and it might take a
long time to upload lots of data. Much research waits to be done in these areas.
Reader innovations
There are several fronts along which research and development in RFID could
contribute to the RFID industry and its applications.
RFID readers today are generic and follow a ‘‘one size fits all’’ paradigm. This
is disadvantageous because readers are overloaded with unnecessary functionality
to serve unknown functions. The cost–benefit analysis of the use of RFID depends
greatly on the cost of readers. There is an emerging need for a larger class of
specialized readers, which we refer to as application-specific readers (ASRs).
Several classes of ASRs are possible.
1. Lightweight WAN-enabled portal readers. Surprisingly, the cost of wiring
RFID readers and connecting them to Ethernet is a significant part of RFID
deployments. There is a need for low-power RFID readers that are, for
example, battery-powered, motion-triggered, and equipped with a 2.5G or 3G
backhaul system. The installation of such a system would be extremely
inexpensive. The research question in the design of a system of this type would
be related to power consumption and performance versus cost.
2. Location-aware mobile readers. Mobile RFID readers will find great use in
RFID systems. Already, several companies are offering mobile reader systems.
However, it is hard to interpret the read events from a mobile reader. A read
event in a backroom has a different context than a read event in a truck. There
exists a need to, on the one hand, make the reader location-aware using, for
example, a local positioning system like a WiFi positioning System (WPS). On
the other hand, there is also a need for a local inference system that interprets
this data and provides the logical hooks to make the reads useful.
3. Unmanned vehicle readers. At the more ambitious end of the spectrum, there
might exist a need for readers that are on unmanned vehicles like robots. This
might occur, for example, in the cataloging of a yard full of containers or in
finding library books [22].
Software
There is great scope for research in software systems for RFID. Several areas
present themselves, but I will discuss one in particular. As discussed in the section
on system software, false positives and false negatives create a challenge in RFID
2 Conversation with David Brock, Ph.D., Laboratory for Manufacturing Productivity, MIT (2006).