From wbell@CS.Cornell.EDU  Mon Oct 29 20:59:57 2001
Return-Path: <wbell@CS.Cornell.EDU>
Received: from postoffice.mail.cornell.edu (postoffice.mail.cornell.edu [132.236.56.7])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9U1xto21491
	for <egs@cs.cornell.edu>; Mon, 29 Oct 2001 20:59:56 -0500 (EST)
Received: from dhcp-190.rover.cornell.edu (dhcp-190.rover.cornell.edu [128.84.24.190])
	by postoffice.mail.cornell.edu (8.9.3/8.9.3) with ESMTP id UAA06958
	for <egs@cs.cornell.edu>; Mon, 29 Oct 2001 20:59:54 -0500 (EST)
Subject: 615 PAPER #40
From: Walter Bell <wbell@CS.Cornell.EDU>
To: egs@CS.Cornell.EDU
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
X-Mailer: Evolution/0.16.99+cvs.2001.10.18.15.19 (Preview Release)
Date: 29 Oct 2001 20:59:32 -0500
Message-Id: <1004407194.1130.18.camel@brute>
Mime-Version: 1.0

40) RADAR: An In-Building RF-Based User Location and Tracking System

RADAR is an analysis of the positional information that can be
determined from triangulating a client's position from two or more
inexpensive wireless base stations. The goal is to use the location
information from these available and inexpensive devices to provide
user location data and tracking for location aware services and
applications. The insight is that most location aware applications do
not require positional applications on the order of precision of
inches, but just on the order of the size of a room. Their intuition
is that reception signal strengths of beacon packets from wireless
base stations are stable enough to provide approximate location
information.

They first examine the problem via empirical means; they choose
several points throughout the environment and determine the signal
strength of each base station. From these sample points, they can
approximate a users location in real time based on the observed signal
strengths of the user. The empirical study yields good results, as
user's positions can be determined to within a few meters. This is a
promising solution for low cost location based systems. The problem
with the empirical system is that it requires a large amount of setup
overhead. One would prefer a non-calibrated system instead of the
highly calibrated empirical system. 

In order to improve on the empirical system, they attempt to model the
system as one of radio propagation, and attempt to find a good
propagation model that models indoor environments well. Using the
Floor Attenuation Model they found a good amount of accuracy without
much complexity. The floor attenuation model accounts for large-scale
path loss and includes an attenuation factor for building floors and
helps mask the effects of obstacles between the sender and the
receiver. From this model, they empirically determined the constants
for their radios and found this model to be similar to the empirical
model in precision, but with significantly less setup overhead.

This paper was very encouraging; it implies that with commonly
available hardware we can build location aware applications without
much added complexity. Their results do imply that most wireless
driver interfaces should provide signal strength information up to the
driver user such that these methods can be used, which most wireless
drivers do not.




From gleason@CS.Cornell.EDU  Tue Oct 30 03:29:49 2001
Return-Path: <gleason@CS.Cornell.EDU>
Received: from postoffice2.mail.cornell.edu (postoffice2.mail.cornell.edu [132.236.56.10])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9U8Tlo01366
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 03:29:47 -0500 (EST)
Received: from cypher.cs.cornell.edu (dhcp-147.rover.cornell.edu [128.84.24.147])
	by postoffice2.mail.cornell.edu (8.9.3/8.9.3) with ESMTP id DAA00176
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 03:29:45 -0500 (EST)
Message-Id: <5.0.2.1.2.20011030031648.00b07d10@postoffice.mail.cornell.edu>
X-Sender: gleason@pop.cs.cornell.edu
X-Mailer: QUALCOMM Windows Eudora Version 5.0.2
Date: Tue, 30 Oct 2001 03:29:32 -0500
To: egs@CS.Cornell.EDU
From: Sunny Gleason <gleason@CS.Cornell.EDU>
Subject: 615 PAPER #40
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed

40) RADAR: An In-Building RF-Based User Location and Tracking System

Despite their lack of originality in naming their project,
the system outlined by Bahl and Padmanabhan is
indeed an exciting application of radio frequency signals
for determining the location of a single user in an
indoor setting. A couple of notable ideas that this paper
presents:

1) Existing wireless LAN hardware has sufficient capabilities
for deploying a user tracking system

2) The directionality of RF LAN equipment must be taken
into account when computing distance information

3) A theoretical model of RF propagation can obtain
comparable results to an empirical method


The authors have developed a user tracking system; Read:
a system for tracking *one* user. It is highly questionable
whether this system would work for any reasonable number
of users, given the wide variation in traffic that 802.11
networks could experience. I tend to doubt the utility of
802.11 signal strength as an indication of anything. Microwave
ovens and 2.4ghz cordless phones are two instant headaches
for this system.

The directionality of antennae in 802.11 pcmcia cards is
a useful result.

It is interesting that they were able to find a theoretical
model that fit their building's RF absorption signature. But,
in general, is it really possible to gain useful effects by
considering only line-of-sight propagation effects?




From papadp@ece.cornell.edu  Tue Oct 30 09:46:58 2001
Return-Path: <papadp@ece.cornell.edu>
Received: from memphis.ece.cornell.edu (memphis.ece.cornell.edu [128.84.81.8])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UEkvo02203
	for <egs@CS.Cornell.EDU>; Tue, 30 Oct 2001 09:46:57 -0500 (EST)
Received: from photon (photon.ece.cornell.edu [128.84.239.166])
	by memphis.ece.cornell.edu (8.11.6/8.11.2) with ESMTP id f9UEknk01199;
	Tue, 30 Oct 2001 09:46:49 -0500
Date: Tue, 30 Oct 2001 09:51:04 -0500 (EST)
From: "Panagiotis (Panos) Papadimitratos" <papadp@ece.cornell.edu>
To: Emin Gun Sirer <egs@CS.Cornell.EDU>
cc: Panagiotis Papadimitratos <papadp@ece.cornell.edu>
Subject: 615 PAPER 40
Message-ID: <Pine.LNX.4.21.0110300911170.23987-100000@photon.ece.cornell.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


Review of: "RADAR: An In-Building RF-based User Location and Tracking
System," by P. Bahl, V.N. Padmanabhan

Panagiotis Papadimitratos papadp@ece.cornell.edu

The proposed system is capable of locating and tracking users (i.e., their
portable devices) inside buildings. Unlike its predecessors that were
based on IR technology, RADAR relies on singal strength measurements and
Wireless LAN COTS, thus reducing the deployment cost. The basic idea is
that the measured (at the mobile node) strength of signals is a
function of the mobile node's location and more than one values measured
from different access points (AP's) can be used to infer an accurate user
location.

First, given a set of BS's, a Radio Map of the space has to be created;
the measurements of the received beacon from each AP within range are
recorded along with the exact location coordinates, or in conjunction with
a inner space RF propagation model. Should a mobile node wish to determine
its position, it looks up the database to determine the closest signal
tuple to the actual measurement and determine the location. The underlying
method is triangulation, i.e., the determination of the direction of a
received RF signal given at least two known points/receivers or the other
way around in RADAR's case (the mobile knows where the AP's are and
receives two or more signals). 

The problems with this approach are the possible estimation errors and
shortcomings with the user tracking, i.e., the continuous update of the
location of a mobile node. The basic limitation lies in that RF
measurements are highly dependent on the ambient transmissions, which may
dynamically change, rendering the Radio Map inconsistent (e.g., contrast
the collection of measurements with a single RADAR-enabled node to the
operational system with numerous stations present). Moreover, the issue of
mobile nodes' location tracking has inherent difficulties, due to the
limited number of measurements received by a certain AP. A more accurate
estimation method is needed in order to remove ambiguities of location,
given a history of measurements; that of course would require some
modeling of the mobility pattern, and could draw from works relevant to
cellular mobile systems.

From eyh5@ee.cornell.edu  Tue Oct 30 10:00:14 2001
Return-Path: <eyh5@ee.cornell.edu>
Received: from memphis.ece.cornell.edu (memphis.ece.cornell.edu [128.84.81.8])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UF0Do03750
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 10:00:13 -0500 (EST)
Received: from james (james.ee.cornell.edu [128.84.236.65])
	by memphis.ece.cornell.edu (8.11.6/8.11.2) with ESMTP id f9UF06k01630
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 10:00:06 -0500
Date: Tue, 30 Oct 2001 09:58:41 -0500 (EST)
From: Edward Hua <eyh5@ece.cornell.edu>
To: egs@CS.Cornell.EDU
Subject: 615 Paper 40
Message-ID: <Pine.GSO.4.21.0110300958210.1227-100000@james.ee.cornell.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

RADAR: An Inbuilding RF-based User Location and Tracking System

Paramvir Bahl and Venkata N. Padmanabhan

This paper introuduces RADAR, a RF-based system for locating and tracking
users inside a building. The basic idea of RADAR is it uses both the
measured and computed signal strength information from multiple receiver
locations to triangulate a user's coordinates. In the authors' experiment
with RADAR, which is carried out on a floor of a building with three
testing base stations (BSs) that are equipped with RADAR, one piece of
information that must be learned a priori is the coordinates of the three
BSs as well as the coordinates of each room on the floor. 

In evaluating the performance of the RADAR system, the authors compare the
data obtained by the empirical method with two other methods: random
selection and strongest bast station selection. The results show that the
empirical method fares better in correctly estimating the user location
than the other two methods. 

In order to reduce the dependence on the empirical data gathered in
real-time, a radio propagation model is proposed that takes advantage of
tehoretically-computed signal strength data, whose operation can be
performed off-line. This alternative method removes some of the complexity
inherent in the empirical method, and its performance suggests that
although there is a little suffering in the obtained data accuracy
compared with the empirical method, it is not significant enough and can
be fully justified by the amount of computational power and time saved
from the latter. 

In the experiment, only one user is studied whose location and movements
are located and tracked. There are no results presented to show how well
RADAR performs with multiple users. One of the concerns is the
interference due to signal strenth measured from a neighboring user that
is mistakenly used to compute the location of the intended user. This
concern is essential in addressing the capacity question. That is, how may
users a network of RADAR-equipped BSs may serve to provide sufficient
location information. This is not discussed in the paper. 
 



From avneesh@csl.cornell.edu  Tue Oct 30 11:03:55 2001
Return-Path: <avneesh@csl.cornell.edu>
Received: from capricorn.ds.csl.cornell.edu (capricorn.csl.cornell.edu [132.236.71.92])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UG3so14335
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 11:03:54 -0500 (EST)
content-class: urn:content-classes:message
Subject: 615 Paper 40
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Date: Tue, 30 Oct 2001 11:05:25 -0500
Message-ID: <97C142C1212ED545B0023A177F5349C4053B22@capricorn.ds.csl.cornell.edu>
X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: 615 Paper 40
Thread-Index: AcFhXK+4MQLCIPQQSo+/LL6SvzHiUg==
From: "Avneesh Bhatnagar" <avneesh@csl.cornell.edu>
To: <egs@CS.Cornell.EDU>
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by sundial.cs.cornell.edu id f9UG3so14335

RADAR: An in-Building RF-based User Location and Tracking service

The RADAR system provides location service, based upon RF signal
strength received from nodes within an RF data network. The RF signal
strength of a transmitter is looked up in a signal strength database to
obtain the accurate transmitter position. The main technique used here
is triangulation based upon measured signal strength from multiple
receivers.

The authors present results based upon an emperical evaluation and a
random method and strongest bast station method. The emperical method
gives the best results. This is however aided by an offline process
which builds a data base of signal strength at a set of fixed receivers.
This data base is then looked up for accurate location discovery.

The problems that are faced with this technique are that RF signals are
affected by ambient signals and mobility might involve decrease in
accuracy. It would have been helpful to see some results with more than
one user which would bring into light another aspect of the system, i.e
how scalable it is. 

From daehyun@csl.cornell.edu  Tue Oct 30 11:10:44 2001
Return-Path: <daehyun@csl.cornell.edu>
Received: from wilkes.csl.cornell.edu (wilkes.csl.cornell.edu [132.236.71.69])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UGAgo15991
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 11:10:42 -0500 (EST)
Received: (from daehyun@localhost)
	by wilkes.csl.cornell.edu (8.9.3/8.9.2) id LAA26163
	for egs@cs.cornell.edu; Tue, 30 Oct 2001 11:10:37 -0500 (EST)
	(envelope-from daehyun)
From: Daehyun Kim <daehyun@csl.cornell.edu>
Message-Id: <200110301610.LAA26163@wilkes.csl.cornell.edu>
Subject: 615 PAPER 40
To: egs@CS.Cornell.EDU
Date: Tue, 30 Oct 2001 11:10:37 -0500 (EST)
X-Mailer: ELM [version 2.4ME+ PL54 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

This paper presented a location support system called RADAR.
RADAR is a radio-frequency based system for locating and tracking users
inside building.

RADAR uses the triangulation technique. Multiple receivers gather user
signal and triangulate the user's coordinates. They use both empirically
and theoretically computed signal strength information.  

They placed three base stations which is equipped with a wireless adaptor.
Mobile hosts broadcast packets periodically. Then, base stations record
signal strength information and determine the location that best matches
the observed signal strength data by triangulation. The algorithm is as
follows;
1. Summarize the signal strength samples using the sample mean.
2. Determine the location and orientation that match a given signal
   strength form 1.
3. compare multiple locations form 2 and pick the one that best match
   the observed signal strength. They named their technique as Nearest
   Neighbor in Signal Space (NNSS).
Their experiments showed that a few meters of precision.

I think RADAR is better than Cricket in several points. It does not
require beacons spread through building and listeners on nodes, so it
does not cost extra equipments. And small number of centralized base
stations might be easier to maintain than lots of beacons. The privacy
problem can also be solved if user nodes compute the location instead
of the base stations.
Of cause, there are also shortcomings. For the accurate location detection,
RF measurement should be performed. And scalability might be worse than
Cricket.

From ranveer@CS.Cornell.EDU  Tue Oct 30 11:26:12 2001
Return-Path: <ranveer@CS.Cornell.EDU>
Received: from exchange.cs.cornell.edu (exchange.cs.cornell.edu [128.84.97.8])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UGQAo18876
	for <egs@popsrv.cs.cornell.edu>; Tue, 30 Oct 2001 11:26:10 -0500 (EST)
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="utf-8"
X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0
Subject: 615 PAPER 40
Date: Tue, 30 Oct 2001 11:26:10 -0500
Message-ID: <706871B20764CD449DB0E8E3D81C4D430213A7BD@opus.cs.cornell.edu>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: 615 PAPER 40
Thread-Index: AcFhX5Wu8UsOVk+pQXO68WDy//TJnA==
From: "Ranveer Chandra" <ranveer@CS.Cornell.EDU>
To: "Emin Gun Sirer" <egs@CS.Cornell.EDU>
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from base64 to 8bit by sundial.cs.cornell.edu id f9UGQAo18876

RADAR: An In-Building RF-based User Location and Tracking System
 
This work presents a novel scheme for location tracking without the use
of extra hardware.   Signal Strength measures at base stations coupled
with triangulation techniques are used to determine a user's location.
In the warming off-line phase, data is collected at the base stations
for a number of locations in the coverage area.  Signal strength in the
real-time phase is used to determine the exact user location.
 
Overall, RADAR does a great job in user location and tracking.  It is a
follow-up work to the active badge system and a precursor to the BAT
system.  It is the first system to move away from IR and use RF signal
strength to come up with reasonably accurate location information.  No
extra hardware is required in this system and this feature greatly adds
to the ease in deployability of RADAR.  
 
RADAR should not be compared with Cricket because of different goals of
location tracking compared to location support of Cricket.  A better
comparison would be with the BAT system.  RADAR compromises accuracy
compared to the BAT system for a lesser cost of deployment.  However,
there are other features of RADAR which need more work.  Firstly, the
warming or training in the off-line phase could be a long and painful
process for accurate real-time information.  An alternative theoretical
model has been worked on, but it needs much more work to give good
results.  Secondly, and more importantly, RADAR would give incorrect
results in the presence of power optimization features.  It works as
long as all the mobile nodes transmit with the same power throughout the
coverage area, and this assumption rules out the possibility of any
power optimization.
 
 

From ramasv@CS.Cornell.EDU  Tue Oct 30 11:29:07 2001
Return-Path: <ramasv@CS.Cornell.EDU>
Received: from exchange.cs.cornell.edu (exchange.cs.cornell.edu [128.84.97.8])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UGT6o19474
	for <egs@popsrv.cs.cornell.edu>; Tue, 30 Oct 2001 11:29:06 -0500 (EST)
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0
Subject: cs615 PAPER 40
Date: Tue, 30 Oct 2001 11:29:05 -0500
Message-ID: <706871B20764CD449DB0E8E3D81C4D4301E7F283@opus.cs.cornell.edu>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: cs615 PAPER 40
Thread-Index: AcFhX/4Y03pfWNBtR9mNBaY5jMlalQ==
From: "Venu Ramasubramanian" <ramasv@CS.Cornell.EDU>
To: "Emin Gun Sirer" <egs@CS.Cornell.EDU>
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by sundial.cs.cornell.edu id f9UGT6o19474

RADAR: An In-Building User Location and Tracking System

	RADAR is a location traking system that uses comparision of
signal stength to pre-determined emperical values and triangulation to
locate and track users in any environment.  This system depends on a few
(3) centralized servers that stores the emperical information and
compares upon request by a node. The emperical values are pre-determined
by measuring signal strngths at different location in the environment.
An interesting point made by the paper relates to variation in signal
strength due to a change in direction of transmission or reception.

	Whenever a location is needed, the signal strength seen at that
point is compared with the table of pre-determined values.  Several
closest distances from the the signal source are considered. The same is
done for 3 different signal sources. A triangulation is then employed to
locate the correct location based on the distance from these sources.
In my opinion comparing with a pre-determined values has several
drawbacks.  First of all, changes in the environment such as new noise
sources (microwave), different temperature could affect the signal
strenght values.  Further it is a big pain to walk across the entire
environment in advance and take sample measurements.  

	An alternative explored in the paper is to use path propagation
models to estimate distances.  But path propagation models are not
accurate and may not be realistic for many environments.  Even though
this paper claims that they could successfully use the WAF model, it may
still not be applicatble in a different environment say with a
metal-lined wall.    

From teifel@csl.cornell.edu  Tue Oct 30 11:41:00 2001
Return-Path: <teifel@csl.cornell.edu>
Received: from disney.csl.cornell.edu (disney.csl.cornell.edu [132.236.71.87])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UGewo22102
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 11:40:58 -0500 (EST)
Received: from localhost (teifel@localhost)
	by disney.csl.cornell.edu (8.11.3/8.9.2) with ESMTP id f9UGeri78485
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 11:40:53 -0500 (EST)
	(envelope-from teifel@disney.csl.cornell.edu)
X-Authentication-Warning: disney.csl.cornell.edu: teifel owned process doing -bs
Date: Tue, 30 Oct 2001 11:40:53 -0500 (EST)
From: "John R. Teifel" <teifel@csl.cornell.edu>
To: <egs@CS.Cornell.EDU>
Subject: 615 PAPER 40
Message-ID: <20011030114009.A65596-100000@disney.csl.cornell.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

RADAR:

This paper describes a location-aware system called RADAR, a RF based
system for locating and tracking users inside buildings.  This system
uses base stations to record, process, and track signal strength from
mobile nodes.  This allows the system to accurately track (within a
few meters) mobile nodes--hence allowing the system to provide
location-dependent services.

Their experimental setup consisted of a few base stations (running
FreeBSD of course, rather than the weaker Linux) and mobile nodes
consisting of laptops.  The targeted domain for RADAR is in-building,
RF domain tracking.  They use an empirical method and a signal
propagation model for calculating a user's location.

This paper was a good start at attempting to provide
location-dependent services for mobile computing networks.  Tracking
mobile nodes using this base-station scheme is a simple, yet effective
idea and is practical and efficient to implement with current
technology--and I expect any such future location-dependent servicing
for mobile nodes will look like the setup in this paper.

From viran@csl.cornell.edu  Tue Oct 30 11:44:50 2001
Return-Path: <viran@csl.cornell.edu>
Received: from moore.csl.cornell.edu (moore.csl.cornell.edu [132.236.71.83])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UGino22379
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 11:44:49 -0500 (EST)
Received: from localhost (viran@localhost)
	by moore.csl.cornell.edu (8.11.3/8.9.2) with ESMTP id f9UGihq07062
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 11:44:44 -0500 (EST)
	(envelope-from viran@moore.csl.cornell.edu)
X-Authentication-Warning: moore.csl.cornell.edu: viran owned process doing -bs
Date: Tue, 30 Oct 2001 11:44:43 -0500 (EST)
From: "Virantha N. Ekanayake" <viran@csl.cornell.edu>
To: <egs@CS.Cornell.EDU>
Subject: 615 Paper 40
Message-ID: <20011030114407.G6932-100000@moore.csl.cornell.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

RADAR is another in-building user location and tracking system, that is
based purely on RF signal strength.  The system consists of mobile hosts
that broadcast pings to base stations that then triangulate the position
of the mobile node.

Unlike Cricket, this system relies on the nodes actively advertising their
presence, thus removing any privacy even if the node is just passing
through the environment without taking part in the network.  The
centralized tracking provides a single point of failure; however this does
allow central monitoring of all users, if that is the required purpose of
the system.  Another downsideis that each node has much higher power
consumption, since instead of just passively listening to beacons like in
Cricket, they now actively broadcast signals constantly, alleviating any
chance to go into a power saving mode. Another disadvantage lies in the
need to take prior measurements of signal strength, thus reducing the
flexibility of the system.


From samar@ece.cornell.edu  Tue Oct 30 12:09:28 2001
Return-Path: <samar@ece.cornell.edu>
Received: from memphis.ece.cornell.edu (memphis.ece.cornell.edu [128.84.81.8])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UH9Ro26153
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 12:09:27 -0500 (EST)
Received: from descartes (descartes.ee.cornell.edu [128.84.236.60])
	by memphis.ece.cornell.edu (8.11.6/8.11.2) with ESMTP id f9UH9Jk05103
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 12:09:19 -0500
Date: Tue, 30 Oct 2001 12:08:28 -0500 (EST)
From: Prince Samar <samar@ece.cornell.edu>
X-Sender: samar@descartes.ee.cornell.edu
To: egs@CS.Cornell.EDU
Subject: 615 PAPER 40
Message-ID: <Pine.GSO.4.21.0110301208020.17228-100000@descartes.ee.cornell.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


40) RADAR: An In-Building RF-based User Location and Tracking System

This paper describes a radio frequency based system for locating and
tracking users inside a building. RADAR depends on a few base-stations to
record/transmit beacons to the mobile nodes. It is based on empirical
strength measurements in the building as well as a simple propagation
model for radio frequencies. The authors find that the empirical, signal
strength measurements to later predict the location of the user is more
effective than using the radio propagation model.

RADAR uses the signal strength measurements at the multiple base-stations
to triangulate the user's coordinates. The triangulation can be done
based on empirically determined signal strengths or one based on a rather
simple radio propagation model. In the former case, signal strength
measurements (which are averaged) are done at many points which
effectively cover the whole area along with the orientation of the user.
Then a look-up table is maintained and a nearest neighbor in signal
strength (NNSS) is found out using the Euclidean measure. Then this
corresponding point is concluded to be the user's location. In the latter
case, signal strength measurements are done at different locations to
find out the parameters of a Wall Attenuation Factor (WAF) model of radio
propagation. WAF is essentially a path loss exponent model with a
corrective factor for attenuation due to walls. Then the location of the
user is predicted based on the received signal strength.

Both these models suffer from some limitations. Before setting up either
of the two model at any place, a significant amount of measurements need
to be performed to construct the data-set or to find the attenuation
model parameters. This may need to be repeated every time a base station
needs to be located. For the empirical data set, data needs to collected
for multiple user orientations at each location. This severely limits the
scalability of RADAR and may increase its cost. Furthermore, many
base-stations may be needed as these signal strength values obtained may
not be unique for all the locations. The author don't seem to have
touched the issue of uniqueness of the signal strength at all (which
probably wasn't a issue in their particular set-up). Also, the authors
claim that the parameters for their wall attenuation propagation model
are similar across base-stations have a lot to do with their particular
physical location plan. This may not hold true if the surroundings of the
base-stations (or even the user for that matter) are markedly different.

Also I feel that the authors could have done better with their radio
propagation model. The authors put the Rayleigh fading model, the Rician
model and their "Wall Attenuation Factor" model at par, which is not
technically correct as the first two are Multipath or short term fading
models (which are used in conjunction with a long term or shadow fading
model) and their WAF model is simply a long term fading model.

From ashish@CS.Cornell.EDU  Tue Oct 30 12:10:24 2001
Return-Path: <ashish@CS.Cornell.EDU>
Received: from cucsloan4 (pit066.cs.cornell.edu [128.84.223.166])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with SMTP id f9UHAMo26238
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 12:10:22 -0500 (EST)
Message-ID: <000f01c16165$c25f4820$a6df5480@cs.cornell.edu>
From: "Ashish Motivala" <ashish@CS.Cornell.EDU>
To: <egs@CS.Cornell.EDU>
Subject: 615 PAPER #40
Date: Tue, 30 Oct 2001 12:10:21 -0500
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000

RADAR: An Inbuilding RF-based User Location and Tracking System

Paramvir Bahl and Venkata N. Padmanabhan

The paper introduces a simple and cost efficient way to identify user
location by triangulating a the position from two or more wireless base
stations. It uses Signal Strenght as a basis for its position match, on 70
points and 4 directions in the space. It determines the location by finding
the Nearest Neighbor in the Search Space using lowest Euclidian Distance as
their metric. They go on to suggest a number of methods to improve and get
better position information

I am unsure of the scalability this location detection scheme. Also using
radio frequency, has it disadvantages, especially in terms of SS and SNR
which will vary drastically everytime in put my taco in the microwave. I
believe that using a couple of metrics together might help.Using machine
learing to identify common locations a user might be in should as be
benefitial as most users are more likely to be at a few places most of the
time. ie A person's location distibution is also more likely to be Zipfian.

From andre@CS.Cornell.EDU  Tue Oct 30 13:02:31 2001
Return-Path: <andre@CS.Cornell.EDU>
Received: from postoffice.mail.cornell.edu (postoffice.mail.cornell.edu [132.236.56.7])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UI2To04125;
	Tue, 30 Oct 2001 13:02:29 -0500 (EST)
Received: from khaffy (d7b102.dialup.cornell.edu [128.253.157.102])
	by postoffice.mail.cornell.edu (8.9.3/8.9.3) with ESMTP id NAA09018;
	Tue, 30 Oct 2001 13:02:27 -0500 (EST)
Received: from andre by khaffy with local (Exim 3.31 #1 (Debian))
	id 15yXd2-00007F-00; Tue, 30 Oct 2001 13:04:24 +0100
Date: Tue, 30 Oct 2001 13:04:24 +0100
From: =?iso-8859-1?Q?Andr=E9?= Allavena <andre@CS.Cornell.EDU>
To: egs@CS.Cornell.EDU
Cc: andre@CS.Cornell.EDU
Subject: 615 PAPER 40
Message-ID: <20011030130424.B404@khaffy>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.3.20i
Sender: =?iso-8859-1?Q?Andr=E9_Allavena?= <andre@CS.Cornell.EDU>


RADAR, An In Buildinng RF User Location and Tracking System.


In this paper, the author describe a location system using the 
following idea: there are bases stations which receive beacons from the
user, and the set of coordiantes (base staion $i$, strength of the
signal received at $i$) is then matched with a set of predetermined
points for which we know the actual physical location.
Put it another way: we know a set of physical point and their values.
Given a set of values, we look to find the point with the closest
values, and assume we are near this physical location.

The set of reference points can be constucted by hand (you input were you
are as well as signal strenght). They call this the empirical model.
Another way to do it is to know the propagation model (the attenuation
model), to be able to take into account the walls, and have a one-to-one
mapping between signal strength and distance from the station. This
gives slightly less good results.

In both cases, the user direction has some importance, and not knowing
it decreases the perofmances. 
In tsead of picking the best match when looking for the location but
averaging on the few best matches can improve the location.


There system needs the space to be covrered by at least two base
stations. The "empirical model" can be quite long to set up. 
Regarding the propagation model, they needeed to mesure the wall
attenuation. There could be some plave where the walls are 
heterogeneous, and there model might be somehow wrong. Also, I am
wondring if the structure of the building (steel vs concrete or
reinforced concrete) wouldmake a difference (the quantity of steel and
its alignement will be quite different).

What about different floors? It might be straightfowrad to add a 3rd
dimension, maybe not

-- 
Andr� Allavena                     (local) 154 A Valentine Place   
�cole Centrale Paris (France)      Ithaca NY 14850 USA
Cornell University (NY)            (permanent) 879 Route de Beausoleil 
PhD in Computer Science            06320 La Turbie FRANCE

From jcb35@cornell.edu  Tue Oct 30 14:01:21 2001
Return-Path: <jcb35@cornell.edu>
Received: from travelers.mail.cornell.edu (travelers.mail.cornell.edu [132.236.56.13])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UJ1Jo12070
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 14:01:19 -0500 (EST)
Received: from travelers.mail.cornell.edu (travelers.mail.cornell.edu [132.236.56.13])
	by travelers.mail.cornell.edu (8.9.3/8.9.3) with SMTP id OAA27243
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 14:01:13 -0500 (EST)
From: jcb35@cornell.edu
Date: Tue, 30 Oct 2001 14:01:13 -0500 (EST)
X-Sender: jcb35@travelers.mail.cornell.edu
To: egs@CS.Cornell.EDU
Subject: 615 PAPER 40
Message-ID: <Pine.SOL.3.91.1011030130944.12409A-100000@travelers.mail.cornell.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


This paper evaluated a method of using triangulation over RF-based 
devices to track a mobile node's location.  The RADAR system uses a 
method based on triangulation based on signal strength to approximate 
location.

Their experiments were carried out on a floor with three base stations 
and one mobile laptop.   By obtaining a tuple of the mean signal strength 
from a few base stations, they can try to accurately determine the 
location based on previous signal measurements by determining which tuple 
the measurements are "closest" to. They also store information about 
orientation, since the direction a wireless unit is facing can 
dramatically affect its signal.

They mention the main limitation of their empirical model is that you 
must know the measurements of many of the tuples before you can be used.  
I could image this would not be reasonable for all cases.

This method also seems highly dependent on the RF environment, since 
anything that would weaken a base station's signal could affect the 
location reading you are given.

From gupta@CS.Cornell.EDU  Tue Oct 30 14:24:50 2001
Return-Path: <gupta@CS.Cornell.EDU>
Received: from ringding.cs.cornell.edu (ringding.cs.cornell.edu [128.84.96.109])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UJOno16102
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 14:24:49 -0500 (EST)
From: Indranil Gupta <gupta@CS.Cornell.EDU>
Received: (from gupta@localhost)
	by ringding.cs.cornell.edu (8.11.3/8.11.3/C-3.2) id f9UJOmh25177
	for egs@cs.cornell.edu; Tue, 30 Oct 2001 14:24:49 -0500 (EST)
Message-Id: <200110301924.f9UJOmh25177@ringding.cs.cornell.edu>
Subject: 615 PAPER 40
To: egs@CS.Cornell.EDU
Date: Tue, 30 Oct 2001 14:24:48 -0500 (EST)
X-Mailer: ELM [version 2.5 PL3]
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

RADAR: an in-building RF-based user location and tracking system, P. Bahl and V.N. Padmanabhan.

Reviewer: Indranil Gupta

This paper presents a location algorithm for tracking stationary or
slow-moving mobile users. The algorithms basically work by using
triangulation depending on the estimated distance to several base
stations, calculated from signal strength information of beacons
transmitted from the base stations. The radio propagation model
attempts to account for signal attenuation due to distance and wall
interference, the only possible major attenuation source within a
building.

The testing set-up maintains a set of 70 X 4 data points for 70 sample
user locations in the buildings, each with four possible
orientations. The location is found by a state space search among
these data points. Multiple nearest neighboring data points can be
used to average out a more accurate user location. Different
orientations can be attempted to hone the space search.

Comments:

Users can have multiple orientations - four is a simplification. Is it
possible that if the user's handheld is not in one of the standard
four positions, the algorithms in the paper might deduce the wrong
location ? Do the authors explain the exact state space search
algorithm used ? If the number of data points is large (say 100000 X 4
instead of 70 X 4) and users are fairly mobile (restricting the time
to do a state space search until the position changes again), a more
efficient search algorithm would be needed.

From gupta@CS.Cornell.EDU  Tue Oct 30 14:25:07 2001
Return-Path: <gupta@CS.Cornell.EDU>
Received: from ringding.cs.cornell.edu (ringding.cs.cornell.edu [128.84.96.109])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9UJP6o16149
	for <egs@cs.cornell.edu>; Tue, 30 Oct 2001 14:25:06 -0500 (EST)
From: Indranil Gupta <gupta@CS.Cornell.EDU>
Received: (from gupta@localhost)
	by ringding.cs.cornell.edu (8.11.3/8.11.3/C-3.2) id f9UJP6r25185
	for egs@cs.cornell.edu; Tue, 30 Oct 2001 14:25:06 -0500 (EST)
Message-Id: <200110301925.f9UJP6r25185@ringding.cs.cornell.edu>
Subject: 615 PAPER 40
To: egs@CS.Cornell.EDU
Date: Tue, 30 Oct 2001 14:25:05 -0500 (EST)
X-Mailer: ELM [version 2.5 PL3]
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

GPS-less low cost outdoor localization for very small devices, N. Bulusu J. Heidemann and D. Estrin.

Reviewer: Indranil Gupta

This paper describes a very simple algorithm to estimate a user
position - this is done by estimating distances from various beacons
(by using a simplistic attenuation model), and estimating the
position as the centroid. Experiments show that the
theoretical error is close to that given by the authors'
implementation.

Comments:

Signal strength differences across beacons, and interference are not
considered while calculating the position. The errors are rather large
(50 % probability for more than a 2 m error on a 10m quadrant - that's
a 20% error with .5 probability). This is ostensibly due to the
simplistic propagation model.

What are the other arguments, besides simplicity of the algorithm, for
using a central beacon (beacons) to measure the _absolute_ position of
each user ? In an ad-hoc routing scenario, it is the _relative_
positions that matter more (it is ok for two neighboring handhelds to
both measure their positions as 2 m s-w of their actual
positions). This could call for a more localized location algorithm -
one that does not require nodes to run through the entire database of
data points whenever the beacon is used to estimate the user
location. Beacons' data is used less frequently, perhaps
periodically. But most of the location update is done only when the
user or one of its neighbors moves - the protocol is more "reactive"
in that sense.

From haom@csl.cornell.edu  Tue Oct 30 16:25:24 2001
Return-Path: <haom@csl.cornell.edu>
Received: from mauchly.csl.cornell.edu (mauchly.csl.cornell.edu [132.236.71.68])
	by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id f9ULPMo13795
	for <egs@CS.Cornell.EDU>; Tue, 30 Oct 2001 16:25:23 -0500 (EST)
Received: from localhost (haom@localhost)
	by mauchly.csl.cornell.edu (8.9.3/8.9.2) with ESMTP id QAA24401
	for <egs@CS.Cornell.EDU>; Tue, 30 Oct 2001 16:25:16 -0500 (EST)
	(envelope-from haom@mauchly.csl.cornell.edu)
X-Authentication-Warning: mauchly.csl.cornell.edu: haom owned process doing -bs
Date: Tue, 30 Oct 2001 16:25:16 -0500 (EST)
From: Ming Hao <haom@csl.cornell.edu>
To: Emin Gun Sirer <egs@CS.Cornell.EDU>
Subject: 615 PAPER 40
In-Reply-To: <200110300148.f9U1mpm03163@zinger.cs.cornell.edu>
Message-ID: <Pine.BSF.4.10.10110301624590.22630-100000@mauchly.csl.cornell.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

RADAR: An In-Building RF-based User Location and Tracking System
by Paramvir Bahl and Venkata N. Padmanabhan

themain idea of this paper is to complement the data networking 
capabilities of RF wireless LANs with accurate user location
and tracking capabilities, thereby enhancing the value of such 
networks.  personally, i think it is a very good approach by taking
advantage of existing wireless LAN. 

the way it works is :
1. there are multiple stations which overlap the cover area.
2. the signal information of each location is collected off-line
3. for the real-time analysis, the nearest neighbor earch is 
   used.



the paper discusses the impact of orientation and multiple neighbors
averaging and teh size of sampling locations.

one of disdvantages of this scheme is that it requires the pre-training
to collect the location-signal strength pair information. it is not 
flexible.

above is the empirical model. there is also a Radio Propagation Model
which does not need to pre-sample the location-signal strengh pair info.
Rayleigh fading model, Rician distribution model and Floor Attenuation 
Factor propagation model. an adapted FAF, called WFA is proposed. the 
benefit of using WFA is that no much measurements are needed and
parameters
of model do not depend on the locations of base stations.


comments. i still think the size of building mustbe very big in order to
make location-aware service make sense. so it would better to for the
paper
to consider a larger building with more base stations. and each station
can 
not cover the whole area.



-ming


Ming Hao
PH.D Candidate, EE		   			 mh97@cornell.edu
Cornell University	    			   Office: (607) 255-0817
Ithaca, NY  14853	                 http://www.ee.cornell.edu/~haom/