Hectic time ahead

This upcoming 2 weeks are going to be fun... 3 tests, Ch 4 & 5, assignments...

At least I no longer have to deal with Industrial Training any further... Confirmed my placement at JPN Kel.

litrev1a

2. Literature Review

(Note that proper literature reviews have yet to be done)

(Pics removed to limit size)

1. Introduction

 

1. Organisation

 

Malaysia Airlines (IATA code: MH) is Malaysia’s national carrier that is among the few airlines that serves all 6 continents on Earth. In June 2009, it carried 1.027 million passengers to and from Malaysia and within the country as well.

 

Among the passengers MH carries some have further requests to be catered, and MH, with its “MH for Malaysian Hospitability”, need to make sure that their requests are fulfilled. But given such a number of passengers that MH carries, management of certain groups of passenger can be difficult.

 

2. Research into Existing System

 

SSR codes are used to denote required service for someone, e.g. UNMR for unaccompanied minors, WCHR for wheelchairs and MOML for Halal meal request (not in the scope of the project) and are displayed in the Passenger Name Record (PNR). As excerpted from Indian Jet Airways’ Unaccompanied Minors page:

                           

The following are the formalities/procedures for the UMNR travel - Domestic:

(…)

The child is handed over to our ground staff at the origin.

The child is escorted by our ground staff and handed over to the In- Flight Executive.

(…)

http://www.jetairways.com/, retrieved 10 August 2009

 

When passengers check in at check-in counters (e.g. at KUL), a ground personnel, either employed by the airlines (flying with MH) or by other airlines (MH handles EgyptAir’s ground service at KUL) or special ground service entity (Emirates’ flights at KUL are handled by KLAS), will be dispatched to escort the passenger to the gate s/he is expected to depart from. The agent is typically expected to be with the passenger at all times until s/he boards the plane, by when the role of taking care to the passenger is handed over to cabin crew.

 

Cabin crew has access to a manifest known as passenger information sheet which contains all passengers’ information, including name, seat number, transit point (where the passenger has previously boarded a flight before, or will take another flight after, or both) and SSRs.

 

Upon arrival, the ground agents also have access to the manifest and know who needs what assistance. The passenger normally will wait until everyone disembarks before an agent escorts the passenger to clear immigration (normally with priority) and retrieve checked luggage until the passenger is reunited with his/her family or gets into a taxi.

 

Currently there are no means of tracking the passengers with SSRs (hereafter called ‘passengers’) centrally. If for some reasons the passenger is dispatched to the wrong gate, or is lost in the process and fail to relocate the passenger in time (e.g. a child that gets too excited and runs around the airport, hiding in somewhere), it can result in failure to send the passenger to the right destination, as demonstrated in Continental Airlines’ PR fiasco when they failed to put two Unaccompanied Minors on the right plane.

 

3. Research into Similar System

 

1. Evacuation Management System by Radiant RFID

 

Hurricane seasons, normally during August and September each year, often bring destructive forces to American coastal cities and displacing many residents, and breaking families up in the process. It is even more traumatic for those who require special attentions, namely those due to disabilities, age, or sickness – if they are lost to medical personnel their medical records may not be able to be retrieved easily.

 

Such situation became apparent during 2005 hurricane season, when Hurricanes Katrina and Rita, arguably among the worst hurricanes ever hit Americans. When Hurricane Rita hit Louisiana’s neighboring states, Hurricane Katrina’s victims, who had yet to return to the devastated homeland such as New Orleans to rebuild, were forced to evacuate to another place once more, causing emergency personnel to lose track of them. It took some time for the personnel to locate them.

 

Radiant RFID, established in 2004 and is located in Austin, Texas, USA, has developed a tracking system using RFID and GPS to track the evacuees due to any disasters, shown in Figure 2.1. The wristband, as shown in Figure 2.2 with its reader, consists of an Alien Technology Higgs 2 EPC Gen 2, passive, preprogrammed, read-only chip and a custom antenna co-developed by Radiant and RCD Technology, is designed to be tear-proof and boasts a range of over a meter on almost every body size. A special portal is custom-designed by Radiant containing Motorola interrogators to capture data from the tag when an evacuee passes through. (http://www.radiantrfid.com, retrieved 10 August 2009)

 

The wristbands are issued to evacuees when they meet at an embarkation center and then board buses with GPS to track the buses. When the evacuees disembark the buses they pass through another portal with RFID reader, which then updates the evacuee’s current location. Meanwhile, the evacuees’ locations can be identified in a central control center. Alternatively, instead of RFID, evacuees can be read with barcodes as well, although as stated by Kenneth Ratton, cofounder of Radiant RFID, the usage of RFID not only lessens human errors, but also speeds up evacuation process.

 

The system was successfully tested in real-life situations during 2008 hurricane season, when Hurricane Gustav and Ike struck American Mainland. 34,800 people were evacuated and tracked using the system. It took just minutes or seconds to track an evacuee in Hurricane Gustav with this system, compared to weeks to find one during Hurricane Rita.

 

(http://www.rfidjournal.com, retrieved 10 August 2009)

 

2. Hong Kong International Airport (HKIA) RFID luggage handling system

 

Aside passenger handling, luggage handling is one of the key element to determine how efficient the airport is. Currently the tags are 1D barcodes that, contrary to popular belief, do not contain destination airport code; they contain 10-digit numeric code, assigned by check-in agents. The tags are affixed to the luggage prior being sent to sorting system, where they will be sorted to their ultimate destination with their tags.

 

The usage of barcodes mean that, in order the tag to be read, it has to be in direct sight, and the tags’ barcodes are not smudged or otherwise made unreadable.  Luggage can frequently gone AWOL, bringing inconveniences to the passenger, who have to live without their supplies, be it temporarily or otherwise, in a foreign place, as well as to the airlines, who have to locate the luggage and pay compensation to the passengers.

 

After 2 years of adoption of RFID-based luggage handling system in HKIA, the system went online in 2005. As shown in Figure 2.3, the luggage that arrive at the airport, either via check-in counters or transfer passengers, are tagged with an RFID-enabled luggage label. Both barcode readers and RFID readers are present before primary sorter, then passing through X-ray scanner and secondary sorter before being tracked with RFID at Lateral. Throughout luggage sorting area there are more than 200 readers, at least 500 antennas, as shown in Figure 2.4, and more than 200 handheld terminals, shown in Figure 2.5, operated with 4W of power level per Hong Kong regulations. The RFID tags are class-0 tags with 96 bits of pre-encoded UID (unique identifier) adopting UHF band of 920-925MHz.

 

Since the adoption of RFID-based luggage handling system, the airport can identify a particular piece of luggage more accurately, with up to 97% of read rate, compared to 80% of accurate read rate using conventional barcode tags. Using RFID luggage tags also enhances security for the airport management and airlines, as well as reducing cost associated with lost luggage.

 

4. Comparisons of Different Systems

 

	HKIA	Radiant RFID	SSRPTS
Tracking System	✓	✓	✓
Real-time information	✓	✓	✓
Privacy with perforated antenna	✗	✗	✓

 

 

 

 

5. Problem Solving Technique

 

1. RFID

 

Radio-frequency identification, or more widely known as RFID, is a system where a reader reads tag(s) via radio frequency waves and interprets the content with a predefined format. The technologies that RFID system uses have been here for a while, but only recently the usage has become widespread that now many are not aware that they are using items that contain RFID tags in it.

 

Various RFID systems are in the market for different purposes and the cost for each system vary greatly, but their functionality, as well the technology employed, are essentially the same.

 

Each RFID tag has the following essential components:

 

1. Antenna – primarily for transmitting and receiving radio frequencies to communicate with the readers. Passive tags also use the antennas to collect energy.

 

2. Integrated Circuit (IC) – An essential part of a tag, its job is to transmit the tag’s unique identifier, and also as a master controller for more complex tags.

 

 

3. Printed Circuit Board (PCB) – Holds the tag together. Depending on the purpose, it can be rigid or flexible, as well as being made of different materials.

Two major classifications of RFID tags, active tags and passive tags are in the market today. An active tag has its own power source, typically battery. Such tags have greater ranges compared to passive counterparts and also can be read through impenetrable materials, but they are larger and more expensive, mostly due to batteries, and the batteries limits its lifespan, depending on its beacon rate, which is the interval of the signal being sent. At up to 8MB, active tags also have larger amount of memory

Passive tags, on the other hand, are cheap to manufacture, much smaller, and therefore, more common on the market. Passive tags collect energy through its antenna. They, however, have comparatively limited range, and may not be read through impenetrable materials. Furthermore, passive tags are notably lower in memory, often not exceeding 64KB.

In order to read and understand the content of RFID tags, appropriate readers are needed for the purpose. Essentially there are at least two components needed for any reader:

1. Antennas are needed to receive and transmit radio frequency. The effectiveness of RFID deployments largely depend on the selection of appropriate antennas.

 

2. Integrated Circuit (IC) board handles necessary information to communicate with the tag. Each board contains a microprocessor, memory and a radio frequency transponder.

 

2. Java

 

Java programming language is an object-oriented programming language that executes its programs in a virtual machine, known as Java Virtual Machine, making the programs written in Java platform-independent. This makes Java a more versatile option when making a program.

 

In addition there are several Java library packages that are made available for users to use on their program, known as Advanced Programming Interface or API. The library is ever-expanding with efforts from programmers throughout the world.

 

An RFID API has been made available at sun-http://rfid.dev.java.net that should be able to provide RFID functions to the programmers.

 

6. Summary

 

As existing system does not offer any tracking system for passengers with SSRs, it is apparent that one should be built to better locate those passengers in near real-time situations. This project’s technological implementations have been studied for best effectiveness and efficiency. RFID set-ups will be used as a measure to track the passengers, while Java is chosen to implement the project and program.

Methodology

3. Methodology

 

1. Introduction

 

Methodology is a systematic and structural procedure that is used as a general guideline for developing a system. It is important that an appropriate methodology flow is adapted to develop a system to meet the requirements as specified.

 

To indicate time allocated to a development phase, Gantt Chart is used to plot the timeline in accordance to the methodology used.

 

2. Methodology choice

 

The system development process utilizes Rational Unified Process (RUP), as shown in Figure 3.1, which was created by Rational Software Corporation, a division of IBM. RUP is an iterative software development process derived from the work on the UML and associated process.

 

Figure 3.1  RUP flowchart (Rational Software)

 

Four Major phases of the process are:

1. Inception, where feasibility tests are conducted to gather requirements and define user requirements. It also assesses risk and cost as well as timeline.
2. Elaboration, where a large majority of the user requirements are defined by creating use-case diagrams, conceptual diagrams and package diagrams.
3. Construction, where the requirements, use-cases and system architectures are taken into account to build the system. It is also where UML diagrams are used.
4. Transition, where the system, documentation and training are delivered to the customer and feedbacks are retrieved.

 

3. Justification

 

RUP is used to develop the project due to the methodology’s characteristic:

• RUP is suitable for real-time systems, which the developing system retrieves data from RFID tags at all time to pinpoint their locations at near-real time.
• The project’s requirements are still not fully understood, and may be changed at any time due to budget changes or technology advancements.
• The system is expected to handle large amount of data.

Waterfall model is not suitable because the model requires its requirements to be fully understood, and is not suitable for real-time data processing. Prototype model is more suitable for online systems and not server-client systems.

 

4. Phases within the chosen methodology

 

Several phases in RUP-based approach in designing the system:

 

1. Inception

There is no known existing system that can track passengers real-time within an airport, and passengers may accidentally get lost in the terminal building. Therefore existing system is examined to determine how a tracking system will it in the existing system without incurring major modification to the system which risks introducing more hidden bugs to the system. Similar, existing RFID-based systems are studied to determine which setup is more appropriate to the system

 

Detailed user requirements about the system are gathered from clients and use-cases are defined.

 

2. Elaboration

Use case diagrams, conceptual diagrams and package diagrams for the system are designed in this phase, and a large part of the system’s requirements are defined more thoroughly.

 

3. Construction

UML diagrams are used in this phase to build and code the project along with the information on developed user requirements, use cases and system architectures. The system will be built as a model basis, since it is impossible to build full-scale tracking systems in a real environment.

 

 

 

 

4. Transition

The system is tested on other students to test its full functionality and retrieve feedback from the testers in order to improve the system as well as to rectify any issues.

 

5. Hardware

 

The system utilizes RFID readers and tags apart from the usual computers and servers.

 

1. RFID readers and tags

In real-life implementations, the system will consist of an array of UHF RFID readers, each with a range of up to 5 meters in radius, while the tags will be class-0 passive tags, embedded into the boarding passes with pre-encoded 96-bit UID with perforated antennas for added privacy and security once the desirability of being able to be read from a distance is no longer wanted.

 

In simulated environment, however, rewritable RFID tags are used to demonstrate the functionality of the system; as the readers only have very limited range of around 5cm, the functionalities will be demonstrated small-scale.

 

6. Software

 

Several software are used to develop the project, to document all related documentations, and to run the system.

 

1. Java Development Tools (JDK)

Developed by Sun Microsystems, Java is among the most widely used object-oriented language JDK is used to implement compilations of the language.

 

2. Rational Rose

Rational Rose is used to create system models, use case diagrams, sequence diagrams, etc.

 

3. MySQL

MySQL is an open-source, multi-platform database system that will be used to store passengers’ records as well as UID for passengers with SSRs.

 

7. System requirements analysis

 

System requirement analysis is critical to system development as it determines if existing hardware/software is capable of the system. The system consists of a central tracking PC, a server for database, and RFID readers. All these are linked up. All PCs and servers in the system run Windows operating systems.

 

8. Summary

 

Appropriate methodology need to be chosen to facilitate project development; in this case, RUP is identified as the most appropriate methodology to be used for the system. A Gantt chart is used to record tasks in each corresponding phase for future references. Tools and software needed to build the system are identified.

Empangan Hijau a la Malaysia

If you haven't heard, our government is mulling a filtering software like that Chinese Green Dam that was withdrawn at the very last minute to be installed on every PCs.

That definitely won't go well among netizens, especially those who believe this is yet another attempt to tighten censorship. While Najib tries to calm netizens down by assuring that the software is intended to filter pr0n sites and nothing more, this image sums their confidence up:

While we are at it, it appears that NSW's anti-pornographic filtering system installed by the state education dept. filters education sites...and leaves those pr0n entries intact... 

User Requirement interview

Just called both MH station at JHB as well as the airport. MH said that I should refer to station manager at the airport, while Senai said the HR dept. should be the one I refer to, both during weekdays office hours.

That proves rather problematic as I have no transportation ( That Proton Wira back in KB that is earmarked for my usage here is being fixed after someone decided it would be a good idea to crash onto it...) and taxi fare to and fro JHB would be like RM60, not to mention I only have Wednesday, Thursday and Friday free after 12pm... Probably it would be easier if I could just do the interview via the phone instead of going to Senai.

Regardless it seems I won't be able to complete Chapter 2 before 13th...

New tasks given

Still having discussions with Cik Hazinah about the PSM meeting. Due next week at 13th :

1. Gantt Chart for PSM1 & 2

2. Chapter 3

3. Chapter 2 - that's gonna be tricky...

Oh and a presentation at 0930 that day.

Privacy Concerns about RFID and Possible Solutions

It can't be denied that RFID is great for tracking items, but due to this attribute it is frequently debated that RFID creates one Orwellian society.

With these tags, some conspiracy theorists (which I always regard them as lunatics) say that manufacturers will have more access to consumers' information, such as shpping habits, usage patterns, and preferences.  They also suggest that government also use it to monitor the citizens.

Regardless of the presence of such conspiracy theorists and their fairy tales, the fact that abuse of this otherwise convenient technology can cause privacy invasion is too hard to ignore. Therefore, IBM came up with one possible solution: Clipped Tag. It behaves exactly like the conventional tags, except the consumers can tear up the antennas along a perforated line so its range can be dramatically reduced from several meters to just a few centimeters. It should be noted the RFID chip is still intact and still contains its data, as long as the reader is close enough to read it.

This is what I plan to implement on the boarding pass - since a portion of the pass is going to be separated, it might as well put antennas on that portion to automatically cut off the antennas when the pass is fed into a reader, leaving the stub with its chip intact to the passenger. This is a crude visualization of the RFID-enabled pass will be:

Application for IT Placement

Just applied for a industrial training position in USMKK's PPKT (AFAIK equivalent to our CICT). Personally I prefer something near my house so that's why I chose this. Let's see what happens next.

Report on RFID

Report on RFID

At A Glance

Radio-frequency identification, or more widely known as RFID, is a system where a reader reads tag(s) via radio frequency waves and interprets the content with a predefined format. The technologies that RFID system uses have been here for a while, but only recently the usage has become widespread that now many are not aware that they are using items that contain RFID tags in it.

Various RFID systems are in the market for different purposes and the cost for each system vary greatly, but their functionality, as well the technology employed, are essentially the same.

History of RFID

In 1888, German physicist Heinrich Rudolf Hertz first proved the existence of electromagnetic radiation; then in 1901 Guglielmo Marconi sent the first wireless long-distance transmission. These created a new era in mankind history.

The first RFID implementation happened in World War II when British military installed “Identification, Friend or Foe” (IFF) on their aircraft, which enabled them to identify the incoming aircraft’s identity – a signal would be transmitted to the aircraft’s transponder which would return an encrypted code. Such technology enabled them to reduce friendly fires among Allied forces. Modern technologies are still essentially the same.

The first RFID patent was awarded by United States Patent and Trademark Office in 1973 to Mario W. Cardullo.[1] Also in the 1970s, Los Alamos National Laboratory was developing a system to track nuclear materials. The system was commercialized in 1980s when the scientists who worked on the project left the laboratory and set up a business.

Between 1999 and 2003, two air interface protocols (Class 1 and Class 0), the Electronic Product Code (EPC) numbering scheme, and a network architecture for looking up data associated on an RFID tag on the Internet, were developed by the Auto-ID Center. Uniform Code Council obtained the license for the technology in 2003, and EPCglobal was created, as a joint venture with EAN International, to commercialize EPC technology. EPCglobal ratified a second-generation standard in December 2004, as a way to anticipate broader adoption.

Basic Usages and Applications of RFID

The simplest form of a RFID system contains three major components:  a tag, a reader, and a predefined format to transfer information. RFID tags transfer information in its memory to the reader via radio frequency waves.

Many of us use RFID-enabled devices daily without realizing its existence. Some of the more notable examples of RFID applications are:

1. 
   Passports – Malaysians got the world’s first biometric passport in 1998, when IRIS Corporation developed the technology using RFID. Many other countries have since begun to issue similar passports.
   


2. 
   Digital Currency - Hong Kong’s Octopus Card, and Japan’s Suica, both based on Sony’s FeliCa RFID smart card system, as well as other equivalent cards, let commuters skip the queues in front of ticket vending machines, and is also a form of payment in convenient stores and at vending machines.
   


3. 
   Toll Payments – Systems like Singapore’s Electronic Road Pricing (ERP) use RFID systems to eliminate the need to stop the vehicle completely to pay the toll charges.
   


4. 
   Libraries – The usage of RFID tags on books as well as other materials facilitate checkouts as well as aid patrons to pinpoint its location.
   


5. 
   Animal Tracking – Pets, livestock, and endangered species can be tagged with RFID chips, and then be tracked to observe its movements as well as to store data such as vaccination records, ownership details, birth date and even temperature.
   

 

RFID Technology

RFID Tags

Several designs of RFID tags are available in the market, each with its own characteristics which make one suitable for a particular application while totally impractical to be used on other circumstances.

Regardless of designs, each RFID tag has the following essential components:

1. 
   Antenna – primarily for transmitting and receiving radio frequencies to communicate with the readers. Passive tags also use the antennas to collect energy.
   


2. 
   Integrated Circuit (IC) – An essential part of a tag, its job is to transmit the tag’s unique identifier, and also as a master controller for more complex tags.
   


3. 
   Printed Circuit Board (PCB) – Holds the tag together. Depending on the purpose, it can be rigid or flexible, as well as being made of different materials.
   

Two major classifications of RFID tags, active tags and passive tags are in the market today. An active tag has its own power source, typically battery. Such tags have greater ranges compared to passive counterparts and also can be read through impenetrable materials, but they are larger and more expensive, mostly due to batteries, and the batteries limits its lifespan, depending on its beacon rate, which is the interval of the signal being sent. At up to 8MB, active tags also have larger amount of memory.

Passive tags, on the other hand, are cheap to manufacture, much smaller, and therefore, more common on the market. Passive tags collect energy through its antenna. They, however, have comparatively limited range, and may not be read through impenetrable materials. Furthermore, passive tags are notably lower in memory, often not exceeding 64KB.

RFID Readers

In order to read and understand the content of RFID tags, appropriate readers are needed for the purpose. Essentially there are at least two components needed for any reader:

1. 
   Antennas are needed to receive and transmit radio frequency. The effectiveness of RFID deployments largely depend on the selection of appropriate antennas.
   


2. 
   Integrated Circuit (IC) board handles necessary information to communicate with the tag. Each board contains a microprocessor, memory and a radio frequency transponder.
   

In some setups it is necessary to have multiple readers set up.

EPC Standards

Electronic Product Code (EPC) is the standard for the RFID system developed by MIT’s Auto-ID Center in 2000. Like Barcode’s Universal Product Code, it identifies an item. However, RFID tags with EPC can identify a particular item, compared to Barcodes, which can only identify an item’s type.

An EPC code contains the following fields:

1. 
   Header – it tells the reader the format of the code.
   


2. 
   EPC Manager Number – it contains the manufacturer’s information.
   


3. 
   Object Class – identifies a class of objects.
   


4. 
   Serial Number – identifies a particular item.
   

Depending on the application, the length of the code can be from 32-bit to 256-bit.

RFID Frequencies

Depending on usages, different RFID frequency range caters different needs for a system. It is particularly important to choose an appropriate frequency range or it may cause interference to other electromagnetic waves such as television signals. Also different countries have different regulations on the RFID’s radio frequencies. The table below summarises frequency ranges available for RFID readers. (Content obtained from http://rfid-handbook.de/rfid/frequencies.html)

Frequency-ranges used for RFID-systems shown with the corresponding field strength and power levels.

Algorithms for RFID

Several algorithms exist for RFID systems to provide required functionalities, but two most important algorithms for RFID are cryptography algorithm and anti-collision algorithm.

While most RFID tags, for example tags found on groceries, do not contain security-sensitive contents other than its EPC codes, some other RFID implementations, for example passports and digital cash systems, require sufficient encryption to protect the user from security risks. Encryptions techniques such as DES, 3DES, AES and proprietary KEELOQ® are used to secure sensitive information that could be exploited to expose privacy and security issues.

When two or more RFID tags are being read simultaneously, or two or more RFID readers read the same tag at the same time, collisions can occur. Such collisions affect accuracy of the RFID data being read, as well as overall performance of the readers.

Readers utilize multiple access communication type to communicate with several tags at the same time. For readers three multi-access procedures exist – Space Division Multiple Access (SDMA), Frequency Division Multiple Access (FDMA) and Time Division Multiple Access (TDMA). SDMA reuses resources such as channel capacity of devices that are spatially separated. The usage of a directional antenna causes the reader to communicate only with a tag in its range. SDMA however is complex and expensive, thereby limits its usefulness. Meanwhile, FDMA divides an RF bandwidth into several smaller frequency bands, enabling tags to communicate in different channels. Like SDMA, FDMA’s usefulness is limited due to its implementation cost. TDMA, on the other hand, divides the available channel capacity into unique time slots that are allocated within each channel. It restricts when and how much a tag can transmit data. TDMA is the most widely implemented reader anti-collision procedures, and is also the most widely used anti-collision for tags as well.[2]

The most notable tag anti-collision algorithm is based on ALOHA, which was first developed in 1970s for a packet radio network at University of Hawaii. Data is sent whenever a station has some. The sender station listens to the broadcast to determine if the transmission is successful or has suffered a collision. If a packet fails to be delivered due to collision, the sender waits for a random period before resending the data. RFID works similarly by listening to the radio before transmitting data. An improvement to ALOHA algorithm, known as slotted ALOHA, divide the time intervals between data transmission, further reducing collision wait time. Further improvement, known as frame-slotted ALOHA, is proposed to alleviate problems associated with collisions when multiple RFID tags are simultaneously present in a reader’s field.[3]

Collision and Multiple Access Protocol

 

Conclusion

RFID is increasingly widespread in daily applications and while it has come concerns, its benefit cannot be denied. However there is also a constant need to improve the existing system to suit the requirements better for one application. Increasing scale of economics to reduce per unit cost is also an important factor in order to make the application more widespread in the future.

---

[1] http://www.rfidjournal.com/article/view/1338/1/129

[2] Banks, Jerry et al. RFID Applied. 2007 ISBN 978-0-471-79365-6

[3] Anti-Collision Algorithm for RFID Tags, Selwyn Piramuthu, Information Systems and Operations Management, University of Florida, 2008, http://tifac.velammal.org/CoMPC/articles/24.pdf