CELLULAR DIGITAL PACKET DATA(CDPD): WHAT MAKES IT RELIABLE?

CONTENTS

• Introduction
  
• Sources of errors
  
• What has to be done to make a network reliable?
  
• Overhead ensures reliability
  
• CDPD vs. other technologies
  
• References
  

In this so-called Age of Information , no one need be reminded of the importance not only of speed but also of accuracy in the storage , retrieval and transmission of data.

SOURCES OF ERRORS

Digital information , by definition, consists of strings of "bits" -- 0's and 1's -- and a physical device , no matter how capably manufactured , may occasionally confuse the two. The most probable reason for this is the low powered transmission of data over long distances.

Mobile communications use radio signals which are subject to eavesdropping. The mobile network is also vulnerable to other unwanted security breaches.

WHAT HAS TO BE DONE TO MAKE A NETWORK RELIABLE?

Fig.1 Data over GSM network

There are some actions that are necessary in order to obtain reliability over a network.

• User Authentication
  

  The procedure which checks if the identity of the subscriber transferred over the radio path corresponds with the details held in the network.

• User Anonymity
  

  Instead of the actual directory telephone number , the International Mobile Subscriber Identity (IMSI) number is used within the network to uniquely identify a mobile subscriber.

• Fraud Prevention
  

  Protection against impersonation of authorised users and fraudulent use of the network is required.

• Protection of user data
  

  All the signals within the network are encrypted and the identification key is never transmitted through the air.This ensures maximum network and data security.

The information needed for the above actions are stored in data bases. The Home Location Register (HLR) stores information relating the subscriber to its network. This includes information for each subscriber on subscription levels , supplementary services and the current or most recently used network and location area. The Authentication Centre (AUC) provides the information to authenticate subscribers using the network , in order to guard against possible fraud , stolen subsciber cards , or unpaid bills. The Visitor Location Register (VLR) stores information about subscription levels , supplementary services and location for a subscriber who is currently in, or has very recently been ,in that area. It may also record whether a subscriber is currently active , thus avoiding delay and unnecessary use of the network in trying to call a switched off terminal.

OVERHEAD ENSURES RELIABILITY

Fig.2 Typical mobile data schematic including mobile data terminal.

You can't always get what you want -- but if you try , sometimes, you just might find you get what you need. Reed and Solomon managed to get a coding system that was based on groups of bits (bytes) rather than individual 0s and 1s.That feature makes Reed-Solomon code particularly good at dealing with "bursts" of errors.

Mathematically, Reed-Solomon codes are based on the arithmetic of finite fields. Indeed, the 1960 paper begins by defining a code as "a mapping from a vector space of dimension m over a finite field K into a vector space of higher dimension over the same field." Starting from a "message" $(a_0, a_1, . . .,a_{m-1})$, where each $a_k$ is an element of the field K, a Reed-Solomon code produces $(P(0),P(g), P(g^2), . . ., P(g^{N-1}))$, where N is the number of elements in K, g is a generator of the (cyclic) group of nonzero elements in K, and P(x) is the polynomial $a_0 + a_1x + . . . + a_{m-1} x^{m-1}$. If N is greater than m, then the values of P overdetermine the polynomial, and the properties of finite fields guarantee that the coefficients of P--i.e., the original message--can be recovered from any m of the values.

Conceptually, the Reed-Solomon code specifies a polynomial by "plotting" a large number of points. And just as the eye can recognize and correct for a couple of "bad" points in what is otherwise clearly a smooth parabola, the Reed-Solomon code can spot incorrect values of P and still recover the original message. A modicum of combinatorial reasoning (and a bit of linear algebra) establishes that this approach can cope with up to s errors, as long as m, the message length, is strictly less than N - 2s.

In today's byte-sized world, for example, it might make sense to let K be the field of degree 8 over $Z_2$, so that each element of K corresponds to a single byte (in computerese, there are four bits to a nibble and two nibbles to a byte). In that case, $N = 2^8 = 256$, and hence messages up to 251 bytes long can be recovered even if two errors occur in transmitting the values $P(0), P(g), . . ., P(g^{255})$. That's a lot better than the 1255 bytes required by the say-everything-five- times approach.

CDPD vs. OTHER TECHNOLOGIES

	Cellular Digital Packet Data (CDPD)	Circuit Switched Cellular	Specialized Mobile Radio (Extended)	Propietary Wireless Data Networks
Speed	best	best	good	good
Security	best	better	good	better
Ubiquity	best	best	good	better
Cost of Service	best	better	better	good
Cost of Deployment	best	best	better	good
Mobility	best	good	better	good
Interoperability	best	good	good	better

References:

"Wireless Data"
IEEE Communications Magazine - January 1995

"Tellabs Wireless"
CDPD vs. Other technologies
http://steinbrecher.com/compare.html

The CDPD Network
John Gallant, Technical Editor, PCSI

Return to contents