Publications & Reports

This is my list of publications. Feel free to email me at dl231@eng.cam.ac.uk with comments about their contents or if you encounter any problems in printing them.

TRIP: a Low-Cost Vision-Based Location System for Ubiquitous Computing
Diego López de Ipiña, Paulo Mendonça and Andy Hopper
Personal and Ubiquitous Computing journal, Springer, vol. 6, no. 3, pp. 206-219, May 2002.

Sentient Computing provides computers with perception so that they can react and provide assistance to user activities. Physical spaces are made sentient when they are wired with networks of sensors capturing context data, which is communicated to computing devices spread through the environment. These devices interpret the information provided and react by performing the actions expected by the user. Among the types of context information provided by sensors, location has proven to be especially useful. Since location is an important context that changes whenever the user moves, a reliable location-tracking system is critical to many sentient applications. However, the sensor technologies used in indoor location tracking are expensive and complex to deploy, configure and maintain. These factors have prevented a wider adoption of Sentient Computing in our living and working spaces. This paper presents TRIP, a low-cost and easily deployable vision-based sensor technology addressing these issues. TRIP employs off-the-shelf hardware (low-cost CCD cameras and PCs) and printable 2-D circular markers for entity identification and location. The usability of TRIP is illustrated through the implementation of several sentient applications.

Download a copy of this journal paper:
Adobe Acrobat .pdf file (5.6 Mb)

Visual Sensing and Middleware Support for Sentient Computing
Diego López de Ipiña
PhD thesis, Cambridge University Engineering Department, January 2002

Offices and homes are increasingly becoming populated with computing devices that are usually networked in some form. This ubiquitous deployment of computing technology is being made possible through the advances experienced in hardware and networking and the more attainable prices. Although computing devices have become more sophisticated and powerful, the way humans use them remains the same, i.e. a desktop-bound model. This gap between the user and the machine has remained unaltered since personal computers were first introduced two decades ago. This research work aims to promote the deployment of a new paradigm of computing, namely Sentient Computing, intended to reduce this gap.

Sentient Computing provides computing systems with awareness of their surrounding environment so that they can change their behaviour according to the observed situation. Rather than forcing the user to explicitly input information, the devices themselves perceive the dynamic world they are part of through sensors. Thus, it is possible to build systems that are attentive to user activities by obtaining even elementary notions of context, e.g. location, identity, proximity or activity. Computer systems and applications, when provided with a certain degree of perception, are made more natural, flexible, and adaptable.

Among the types of context information provided by sensors, location has proven to be especially useful. Since location is an important context that changes whenever the user moves, a reliable location-tracking system is critical to many context-aware applications. However, the sensor technologies conventionally used in indoor location tracking are expensive and complex to deploy, configure and maintain. These factors have prevented a wider adoption of Sentient Computing in our living and working spaces. This dissertation presents a low-cost and easily deployable vision-based sensor technology, which employs off-the-shelf hardware (web-cams and PCs) and printable circular markers for the identification and accurate 3D coordinate location and orientation of tagged entities. Users with a web-cam attached to their PCs simply need to install the software of this sensor in order to provide visual awareness to their systems.

Subsequently, a set of programming abstractions for Sentient Computing are proposed that enable the developer to simply concentrate on the specification of the context needs and the implementation of the reactive behaviour of a particular sentient system. The details dealing with access to the raw data from sensors and its interpretation into the high-level context needs of applications are made transparent to the developer by the set of abstractions defined. Moreover, common functionality within sentient applications, such as the correlation of event-based context notifications and the activation, migration and deactivation of services as a consequence of sensed situations, is supported by network accessible middleware services. The thesis of this dissertation is that the set of technologies suggested for sentient systems, both in the sensing and infrastructure support areas, will enable their easier development and, more importantly, their later deployment even in spaces where before this was infeasible, such as our homes or workplaces.

Download a copy of this dissertation:
Adobe Acrobat .pdf file (3.5 Mb)

An ECA Rule-Matching Service for Simpler Development of Reactive Applications (HTML)
Diego López de Ipiña and Eleftheria Katsiri
Presented at the Work in Progress session of Middleware 2001, Heidelberg, Germany, 12-16 November 2001. Published as a supplement to the Proceedings of Middleware 2001 at IEEE Distributed Systems Online, Vol. 2, No. 7, November 2001

This paper describes the implementation of a CORBA-based Event-Condition-Action (ECA) Rule Matching Service. This service aims to add composite event management capabilities to the standard CORBA Notification Service. The originality of this work resides on having leveraged on the pattern matching capabilities of a production systems programming language, i.e. CLIPS, for the implementation of a distributed event correlation and aggregation process. A powerful yet simple ECA rule specification language is defined that hides the inconvenient syntax of the underlying CLIPS language and extends it with convenient constructions for event correlation. The ECA Service main capabilities are to enable: (1) composite event matching, (2) new event aggregation and to (3) trigger a set of pre-defined and arbitrary actions as a consequence of a matched situation. This service highly simplifies the development of reactive applications by alleviating the programmer from the implementation of complex composite event handling mechanisms. The potential of this service is illustrated by its use in the implementation of a context-aware application.

Download an extended version of this paper:
Adobe Acrobat .pdf file (57K)

Sentient Computing gives perception to computing systems so that they can detect, interpret and respond to changing aspects of user contexts. The location attribute of a user's context is of special interest because it makes human-computer interactions more natural. In the last few years, several sophisticated indoor location technologies, which can track user whereabouts, have been developed. However, they are yet to be widely adopted because of their high cost and complexities in deployment, configuration and maintenance. This paper describes a novel vision-based software location system, known as TRIP, whose low-cost, off-the-shelf hardware requirements and easy deployment features overcome other systems' limitations. Nevertheless, in order to foster the deployment of "sentient spaces" that bring services to users wherever they are or about to move to, a location system must also be accompanied by the middleware to facilitate user-bound software service activation, migration and deactivation. LocALE addresses this issue by providing a CORBA-based solution that deals with heterogeneous object lifecycle and location control. Some distributed applications combining TRIP's and LocALE's capabilities are presented to demonstrate that Sentient Computing can be made readily available for everyone and everywhere.

Download full paper:
Adobe Acrobat .pdf file (1.6M)
Postscript - A4 compressed .ps.gz file (0.8M)

Sentient Spaces are perceptive physical spaces that aid user activities with computer provided services. Context information, captured through networks of sensors, is communicated to computing devices, embedded in the environment, that interpret the current situation observed and, as a consequence, trigger adequate services for the user. Unfortunately, the sensor technologies used in context capture (in special user location) are usually too expensive and complex to deploy, configure and maintain. This has prevented a wider adoption of the so-called Sentient Computing paradigm in our living or working spaces. We present, TRIP, a novel vision-based sensor that uses the combination of off-the-shelf hardware (video cameras and PCs) and printable circular markers for entity identification and location. This more convenient sensing device is accompanied by a set of programming abstractions and middleware services, named SIF, that make the development of sentient systems and their later deployment a less cumbersome and easier process.

Download full paper:
Adobe Acrobat .pdf file (6.1M)
Postscript - A4 compressed .ps.gz file (1.5M)

The LocALE (Location-Aware Lifecycle Environment) framework provides a simple management interface for controlling the lifecycle of CORBA distributed objects. It supports mechanisms for the remote construction, movement, removal and recovery of heterogeneous software objects in a location domain, i.e. a group of hosts on a network within a given physical area. Client applications use LocALE to intelligently control their required services' location and relocation in the network. LocALE offers load-balancing, automatic activation, and fault-tolerance facilities for the services whose lifecycles it controls. It provides the middleware necessary for the efficient implementation of location-aware mobile applications in richly equipped network environments. LocALE's infrastructure has been tested with the development of several follow-me applications that dynamically move with their users as they change location. For illustration, two of these follow-me LocALE-enabled applications are described.

Download full paper:
Adobe Acrobat .pdf file (73K)
Postscript - A4 compressed .ps.gz file (190K)

TRIP (Target Recognition using Image Processing) is a novel vision-based sensor system that uses a combination of visual markers (2-D circular barcode tags, or ringcodes) and conventional video cameras to identify tagged objects in the field of view. A CORBA-based distributed component architecture called Sentient Information Framework has been devised to efficiently manage and distribute to applications the sensor data obtained both from TRIP and other sentient technologies. This paper describes the implementation, in Python, of two components for this framework; a TRIP Directory Service mapping ringcode identifiers to attributes and a context abstractor type component insulating low level details of sensor data acquisition and interpretation from an application. These case studies will reflect the potential of Python and CORBA, assisted by its Event Notification Services, as an ideal technology combination for the rapid development and efficient gluing of heterogeneous distributed software components.

Download full paper:
Adobe Acrobat .pdf file (800K)
Postscript - A4 compressed .ps.gz file (263K)

This report describes the development of a novel sensor technology named TRIP (Target Recognition using Image Processing) that uses the combination of visual markers (2-D circular bar code tags) and video cameras to automatically identify tagged real word objects in the field of view. A CORBA event-based distributed component architecture employed to manage and distribute to applications the sensorial data provided by TRIP and an example application that benefits from it are also overviewed. Finally, extensions to this sensor technology and potential applications are proposed and a schedule is given of how the author will proceed to complete this research in his remaining two years of PhD.

Download full report:
Adobe Acrobat .pdf file (5.3M)
Postscript - A4 compressed .ps.gz file (1.5M)

Recently, efforts have been made to incorporate full-motion, high quality interactive Video into Distance Learning Systems. Initial results have shown that Distance Learning can benefit greatly when good quality video playback is achieved. Development of this type of systems, however, poses great challenges: digital video not only requires significantly more storage space and transmission bandwidth than traditional data services, it must be delivered in time, i.e. isochronously, for continuous playback. So far, existing Distance Learning Systems have always been developed using proprietary software and constrained to only one client platform. This work addresses a software openness approach to video-based Interactive Distance Learning development, using the latest beta release of an Oracle Server System and up-to-date Java technology. The project analyses the trade-offs taken and difficulties encountered to produce an acceptable performance multi-platform solution.

Download full thesis:

Adobe Acrobat .pdf file (2M)
Postscript - A4 compressed .ps.gz file (3.2M)

Download Interactive Video Client Java Code:
Click here to obtain source code.