Document Sample
p1 Powered By Docstoc

Gil, this is what I have as of Monday at 5pm. This is generally a
mess, I will have time tomorrow to finish it.

Indoor Positioning Systems

1. Introduction
        We are developing a local positioning sytem (LPS) that is capable of operating
indoors. The advent of Global Positioning Systems (GPS) has revolutionized navigation.
However, the GPS network does not function inside most structures, i.e. office buildings,
factories, etc. The reason is that GPS uses radio waves to operate. However, radio waves
from GPS satellites, which are a form of electromagnetic radiation, do not propagate in
very well inside a building. The main reason for this deficiency is that GPS signals are
low power (-130dBm outside, sea level), and are so severely attenuated inside buildings
that GPS receivers cannot detect them. Furthermore, even if the satellite’s carrier signal
could be detected, the attenuation is so large that the GPS receiver could not lock onto the
modulation due to prohibitively low signal to noise ratios. The main factors that
contribute to the attenuation of EM inside buildings can be traced to the conducting and
dielectric properties of building walls. If the GPS system could greatly increase the power
output from their orbiting satellites, indoor positioning would be much more likely.
However, there are two fundamental reasons why this is highly unlikely. One is that
orbiting satellites can not output significant power without being quickly depleted. The
second reason is that GPS operates with continuous wave (CW) propagation, and
transmitting high power continuously could have harmful effects on people and the

        We are developing technology that enables indoor positioning. The indoor
positioning technology is based on a new method for positioning. Furthermore, the new
technology is being developed with the aid of a unique CAD tool for electromagnetics.
This new EM CAD tool, was developed by our design team, and is not available
anywhere else. The LPS system will be able to transmit much more power to the
receiver than the satellite system. In addition, instead of transmitting continuously with
CW, we employ pulsed power with EM waves propagating for time scales that are on the
order of nanoseconds. This pulsed technique obviates the need for the receiver to lock
onto the modulation scheme of a transmitter. It also avoids the potential problem of
harming people and the environment because the EM signal is only present for
nanoseconds, and therefore transmits negligible total energy.
        To achieve this new LPS technology for indoor positioning, we are combining our
expertise in three fields: Radio Frequency Very Large Scale Integration (RF VLSI)
Electronics; Electromagnetics (EM); and Statistics. Our experience in RF VLSI facilitates
the design and fabrication of the necessary microelectronics hardware. The team’s
expertise in electromagnetics will enable the optimized design and fabrication of
antennas specific to the LPS applications, as well as the enhancement of our new state of
the art EM modeling code for predicting the propagation of RF signals inside structures.
Finally, by drawing on our work on statistics, we are developing techniques that extract
the correct location information based on an overly determined data set.

The products we plan to bring to market include:
    A local distance determination system capable of indoor operation.
    A local absolute positioning system capable of indoor operation.
    Software for predicting the path of electromagnetic wave propagation with
       applications to positioning.

Description of Proposed Solution

2.1 Local Positioning System: Underlying Theory
       Below we describe the basic operating principle of our LPS system. We start by
explaining how we use electronics and RF to measure absolute distance. We then explain
how we extend the principle to measure precise location.

2.1. Measuring Absolute Distance
        To measure absolute distance we use ultra-fast clocks along with the speed of
light. Inexpensive ultra-fast clocks are now possible to build as a result of the
microelectronic revolution. Complementary Metal Oxide Semiconductor (CMOS)
Transistors are the basic building blocks of most modern integrated circuits (chips).
Technology has moved so quickly that it is now possible to routinely build chips with
millions of transistors that have critical dimensions of less than 0.2 microns. In addition
to packing a large number of transistors on a chip, their small size allows these basic
building blocks to operate on time scales of less than 0.1 nanosecond. It is now possible
for even the small business to build chips using these state-of-the-art transistors. We
design our circuit, and then contract out the fabrication to the manufacturer. (The
manufacturer we use is the MOSIS facility, which specializes in small volumes.) This
enables even a small business to develop products based on the most modern technology.
        We have used ultra-small transistors to build ultra-fast clocks. In fact, with such
small devices we have designed, and had electronic clocks fabricated that operate so fast
that we can use them to measure the speed of light. For example, we have designed
electronic clocks that can measure times as small as 0.1 nanosecond. By knowing that the
speed of light is 3 X 1010 cm/sec, we can use our clocks, in conjunction with
electromagnetic wave propagation, to measure distances with a resolution of 3cm [(3 X
1010 cm/sec)(1 X 10 –10 sec )=3cm].
        To understand how we measure distances using fast clocks and electromagnetic
(EM) waves, consider the following. At location ‘A’ we have a transceiver that is capable
of sending and receiving electromagnetic signals. Connected to the transceiver is a very
high frequency clock that is operating at a known frequency of say 10GHz. At location
‘B’ we have another transceiver. To measure the distance between points A and B, the
transceiver at point A sends an EM signal to B, at the same time the clock at A starts.
After a finite amount of time, transceiver B receives the signal and transmits it back to A.
When A receives the signal the number of periods on the clock is recorded, which is the
time it has required for the EM wave to go to from A to B and back to A. By multiplying
this time by the speed of light, we can determine the distance between A and B. Intrinsic
delays due to the electronics response times for the electronics will be easily measured
and calibrated.

2.2. Measuring Precise Location
        Suppose we want to measure the location of a point B. We achieve this by
extending the above methodology by using two more transceivers. We place transceivers
A1, A2 and A3 at three known locations. Each transceiver has its own clock. Using the
algorithm discussed above, we can find the distance between point B and A1, A2, A3. By
knowing the distance between B and the three positioned transceivers, a simple geometric
relationship will give the precise location of point B relative to A1, A2 or A3. (This is
analogous to the GPS triangulation methodology.)

3. Core Hardware Design
        The LPS system will be composed base stations and personal LPS devices. The
base stations will be mobile, but their positions will remain fixed once they are put into
service for a specific LPS event. The personal LPS devices are mobile and worn by safety
workers (firemen, or any other individual that we want to track). The personal LPS
device is mainly a transceiver. The base station consists of a transceiver, a GHz clock, a
counter, and digital logic for converting counts to distance. (Of course there is additional
circuitry identifying individual personnel. This is described below.)
        The personal LPS transceiver consists of a low noise input amplifier that is tuned
to a specific very narrow band frequency range. The output signal is generated by voltage
controlled oscillator that is mixed with a square wave of very low duty cycle. The
resulting output will be an amplitude shift key modulated (ASK) carrier wave (pulsed
sinusoidal) which is then fed into an RF amplifier to achieve power levels on the order of
several watts. The output of the amplifier will then be impedance matched to drive the
output antenna.

        The LPS system components include contain transceivers, counters and antennas
as their basic components. We will develop two different designs. The first system we
develop will use already existing integrated circuits. A second system, using on our own
chips that are tailored specifically for the LPS system, will subsequently be developed.
By developing our own IC’s we will optimize

       We will first design these transceiver chips, and then have them fabricated using
the MOSIS chip fabrication facility. Our transceivers will be designed and fabricated
using CMOS technology.

       The transceivers will consist of a receiver and a transmitter. The transmitter will
use an generated with a frequency synthesizer that is based on a phase-locked loop. The
transmitter output will be a tuned, common source-type power amp matched to a 50ohm
patch antenna. We will operate at several frequencies. These frequencies will be in the
ISM bands and are likely to include 433MHz, 900MHz, and 2.47GHz. (Below we
describe that our prototype operates at 433MHz.) The advantage of using high
frequencies is that it allows use of passive components that are small, and thus pocket
size transceivers. This frequency is accessible using inexpensive CMOS components
from a 0.25micron feature length process, available through MOSIS. The basic design of
our transceivers is illustrated in Fig. 1, which is a layout of an FSK transmitter chip that
we already have had fabricated using MOSIS.
The receiver section will also be based on the phase-locked loop (PLL) topology. The
input stage will consist of a low noise tuned amplifier, which feeds the PLL. The PLL
will drive a voltage-controlled oscillator that will be used to demodulate the FSK signal.
The high speed clock, used to calculate the time required for signals to travel, will be a
will be a three stage ring oscillator that inputs an asynchronous counter. The clock
frequency will be approximately 10GHz in our prototype. Into the chip will also be
designed the digital circuit that will convert the clock values into distances and location.
The circuits will be designed with the aid of the circuit simulator SPICE, and laid-out
using the Cadence IC development software. We have recently designed several test
chips using MOSIS to establish design parameters and fabricated several of our circuit
building blocks, including PLLs, clocks and counters[1-3].
        We are developing a lower frequency prototype of this system using off-the-shelf
components. Schematics and circuit board layouts for this system are shown in Figures 2
and 3.

2.4. CAD with State-of-the-Art Electromagnetic Modeling

To help determine the frequency, polarization, and power level of the EM signals, along
with base station location and propagation within walled structures we have developed a
state of the art electromagnetic wave propagation computer aided design tool (CAD). Our
novel CAD tool predicts the velocity and power of an electromagnetic RF pulse as it
propagates from a base station to the receiver and back. In addition, the CAD tool can tell
us if there is any deviation from the straight line path of propagation.

Accurate modeling of modern signal propagation inside structures usually requires a full-
wave solution to Maxwell’s equations. However, such a solution is difficult because our
system uses EM waves of very short duration. Therefore, frequency domain analyses are
not appropriate for our applications, so we must rely on time-domain numerical solutions
to Maxwell’s equations. However, Maxwell equation Finite-Difference-Time-Domain
(FDTD) conventional solvers employ methods are limited by the Courant condition. This
restriction requires very small time steps, and therefore prohibitively long simulation
times are required to analyze the details of EM propagation inside buildings. To
overcome this problem, we have developed a new state-of-the-art simulator that uses the
Alternating-Direction-Implicit (ADI) method [1, 2]. In this new FDTD-ADI method.
Maxwell’s equations are discretized with the electric and magnetic fields on different
grids [1, 2]. By manipulating Maxwell’s equations, we transform the differential
equations to a system of tri-diagonal algebraic equations. Each matrix of the system
corresponds to one specific dimension [1, 2]. We then solve the tri-diagonal systems at
each time step for the EM fields in 3D.
        We have developed this software over the past two years. The software is
extremely well suited to model the electromagnetic wave propagation for our location
system because it efficiently operates in the time domain. We have already presented its
capabilities at two different conferences, where it was very well received[REF].
        to calculate distances, we will power levels necessary in order to send signals into
buildings we Measuring distance and location out of doors has less chance of being
complicated by multi-path possibilities. However, inside structures multi-path
propagation of EM waves is likely to occur. In any case, the methodologies we describe
below for discerning the direct path can be applied to internal as well as external
environments. To account for multi-path propagation we will combine our expertise in
VLSI circuit design with our recent work in modeling electromagnetic wave propagation.
By understanding the details of EM wave propagation, we expect to be able to predict the
path of the EM wave between the various location transceivers. To achieve this we will
model the propagation of EM waves between transceivers in edifices composed of
standard building materials and designs. Modeling the propagation of EM waves is
achieved by solving Maxwell’s equations, which are the fundamental mathematical
equations that describe electromagnetism. Maxwell’s equations are a system of four 3-
dimensional, time-dependent partial differential equations. The solution of these
equations describes the characteristics of electric and magnetic fields. These equations
are very complicated, and thus especially difficult to solve. In fact, before the advent of
powerful computers, solutions of these equations were only obtained for the simplest
cases. We have recently developed a new numerical method for solving these equations
for investigating EM effects in computer chips[4]. We plan to extend this method to
modeling EM wave propagation that occurs between transceivers both inside and outside
buildings. The new method employs the alternating direction implicit (ADI) technique.
This technique has the advantage of being able to resolve both large and small objects
simultaneously. This will enable us to simulate the propagation of EM waves between
positioning transceivers, and therefore help determine the path our EM waves are taking.
Once we have adapted our computer chip field solvers for modeling electric fields inside
structures, we plan to simulate wave propagation inside typical buildings to calibrate the
simulator, as well as determine the optimal locations for our stationary transceivers.

Antenna Design
Type of antenna for base station:
Major considerations:
   1. Directivity. Since the base station is intended to be mobile, good efficiency can be
       achieved by using directional antennas that span half-space.

   2. Polarization. Circular or Elliptic polarized field has advantages over linearly
      polarized field in that it can penetrate through fog, moisture, or other gases that
      are potentially present in the fire scene.
   3. Profile: Wire vs. Surface patches. Microstrip patch antennas, which come in a
      wide variety of shape are ideal for such application because of their low profile,
      cost effectiveness, and ease of manufacturing. Wire antennas, on the other hand,
      have an en extended profile that allows for increased efficiency but at the cost of
      volume. Microstrip antennas will be the first choice. Increasing the efficiency of
      these antennas can be made possible by increasing thickness of the substrate and
      using low-loss material.

   4. Miniaturization. Microstrip patch antennas are most efficient and effective when
      they are constructed on a ground plane which is ideal, i.e., extends to infinity.
      Because of the portability of the base station (small size), we intend to use the
      novel concept of high-impedance surface (HIS) to produce an effective ground
      plane that significantly diminishes reflections from the edges of a finite (small)
      ground plane. The HIS also improves the efficiency of the antenna and the
      matching potential as it eliminates the ripples in the input impedance.

The personal antenna, which will be placed on the protective suite of the fireman (or his
protective helmet), needs to be an isotropic radiator, with equal efficiency in all
directions. Several options will be considered such as a fat monopole to maximize
efficiency. Patch antennas can be considered as an option. Initially, we intend to use off-
the-shelf antennas. However, in-house design of such antennas is inevitable, as such
antennas need to be mechanically and thermally robust while not sacrificing electrical
(radiated) performance). We intended to investigate the effect of coatings on the antenna

Antenna testing will be carried out in an anechoic chamber with sufficient absorption
range over the frequency band of interest. The antennas will be characterized using a
vector network analyzer, and the radiation patterns can be calculated using a spectrum
analyzer. Several test antennas covering the frequency range 30MHz to 5GHz are
available for pattern measurements.


Possible Alternative Approaches (For MIPS only, response to reviewer)
        We have also explored alternative approaches to locate lost safety workers inside
buildings. Our background investigation and market analysis indicate that there are two
possible approaches which may compete with, but are more likely to complement, with
our own method. The first is an audio alarm. A downed fireman my activate an audio
alarm that can lead other safety personnel to the general location. Of course this
approach has applications. However, in the chaotic atmosphere of a conflagration such
an alarm may not be discernable. In addition, the alarm itself does not provide the
identity of the victim. In addition, the victim may have been incapacitated, and be unable
to activate the alarm. Finally, there may be situations where you do not want everyone to
know the location of the downed personnel, so sounding an audio alarm would not be
appropriate under such circumstances. In any case, this approach is fairly simple so it
could be easily added to complement our own LPS design.
         The other possible approach is to use a GPS receiver that is so sensitive that it is
able to function indoors. Motorola Corporation has announced that it plans to bring a new
GPS receiver IC to the market. the MG4000. The company claims that the sensitivity of
the new MG4000 chip will be as much at -153dBm. The outdoor power of the GPS
satellite signal when is reaches the earth is approximately -130dBm. The signal is
reduced by another 20dBm once it enters a single story building, and attenuates another
10dBm for each additional building level[REF Saunders]. The conclusion from this
information is that this nascent Motorola chip should be able to operate in small (one and
perhaps two story structures). We therefore plan to use this chip to complement our
proposed system. We will build a system where the person (or object) whose position we
want to detect, wears a circuit that contains the single chip GPS receiver. On the board
will also be a transmitter that takes this digital word output of the GPS receiver, and
inputs it into a transmitter. The location of the transmitter will then be sent to a base
station receiver.

To develop this technology we will use transceiver multi-chip modules produced by Linx
Corporation. These modules are available in 20 pin dual in-line packages, and are capable
of transmitting digital words of up to 10 bits. We have already prototyped transmitter-
receiver systems based on the Linx module. The major limitation of the Linx module is
that it operates at a frequency of 433MHz. While this frequency seems relatively large, it
is often not sufficiently high to provide pocket size transceivers for all applications. After
prototyping the system with the Linx transmitter module, we plan to raise the operating
frequencies using our own transceiver systems based on the PLL-FSK technology we
described above.

Z. Dilli, and N. Goldsman, MOSIS Design number 65046; Design name: ringosc05;
Technology: SCN3ME\_SUBM, lambda = 0.3; (An oscillator-counter chip test chip)

Z. Dilli and N. Goldsman, MOSIS Design number                 64639; Design name:
diginterf;Technology: SCNA, lambda = 0.8; (An oscillator-counter test chip) 2002.

Y. Bai and N. Goldsman, MOSIS Design number: 65008; Design name: FSK;
Technology: SCN3ME\_SUBM, lambda = 0.3; (An FSK transmitter test chip) 2002.

X. Shao, N. Goldsman, O. Ramahi, P. N. Guzdar, A New Method for Simulation of On-
Chip Interconnects and Substrate Currents with 3D Alternating-Direction-Implicit (ADI)
Maxwell Equation Solver. To be published in 2003 International Conference on
Simulation of Semiconductor Processes and Devices.
Fig. 1: Layout of microchip FSK transmitter we have designed and developed when we reach the
  stage of prototyping the location position system using our own integrated circuits. The actual
                                                            dimensions of the chip are 2 X 2 mm2
      Fig 2: Schematic and Printed Circuit Board layout of our prototype distance
determination circuit. This circuit is to be worn by the person/object whose distance is to
        be determined. The actual dimensions are approximately 3 x 2 inches2.
Fig. 3: Schematic and Printed Circuit Board layout of base station for determining
                 distance. The actual dimensions are approximately 2 x 2 inches .

Shared By: