... Internet¹

. In this book we use the convention of ("big-I") "Internet" meaning the worldwide interconnection of networks. We use ("little-i") "internet" to mean either an internetwork (collection of networks) or a protocol suite (typified by TCP/IP).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 2000.²

. Source: Economic and Management Consultants, Inc., estimate [EMCI95].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... channel³

. "Channel" is one of those words whose meaning varies with the context and the level of "source" and "destination." It could consist of a physical medium or a logical data path.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ear.⁴

. Of course, many messages are conveyed more or less explicitly in the form of body language, in which the channel is the visual medium of electromagnetic radiation commonly known as reflected light. Difficulties arise when these visual messages conflict with the verbal messages they accompany.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... message.⁵

. Correct message reception is one thing, agreement with the message content is another thing entirely. Cultural variances in acknowledgment responses have often led to difficulties in international relationships between people.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... unacknowledged.⁶

. The primary reason for using broadcast and multicast is to gain channel efficiency by communicating with multiple entities via a single common message. However, the complexity of matching acknowledgments with intended message recipients at the source quickly overrides any efficiencies gained in a reliable multicast scenario. Depending on the number of channels, the number of members in the multicast group, etc., a set of unicast messages might be equally efficient for reliable communications. It really depends on where you want to handle reliability-in the networking technology transporting the message or in the application itself. We'll talk more about reliability later.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... bps⁷

. Channel bandwidth is most often expressed in thousands of bits or kilobits per second (kbps) and in millions of bits or megabits per second (Mbps).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... BER⁸

. Channel quality is also measured in terms of block error rate (BLER) and sometimes packet error rate (PER).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... layer.⁹

. The protocols at each layer come from one of essentially three sources: the ISO, the "Internet world" or proprietary sources, such as Novell. The political and philosophical differences between these sources run deep. None of these sources of protocols definitions is without flaw. We will try to maintain a balanced presentation, while acknowledging that the current dominating influence (at least in North America) is what we refer to as the "Internet world."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... envelope.¹⁰

. Given the postal system, this is clearly an example of a connectionless protocol.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (IP)(RFC¹¹

. Internet standards are defined by means of so-called "Requests for Comment" or RFCs, which are numbered sequentially.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... header¹²

. Formally called Protocol Control Information or PCI.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... wirelessness^1.1

. "Wirelessness" refers to the use of a wireless media, most typically radio frequency (RF) systems.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Gary^1.2

. Gary is a true road warrior we know who regularly logs over 300 nights on the road per year. Someone who really needs free airline mileage awards.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Pat^1.3

. Pat is a good name for an administrative assistant example because it avoids all sexist stereotyping. Kim, Chris and Sandy also work in this respect.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... mobility.^1.4

. For now we shall not consider real-world concerns like charging for services,etc.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... provider.^1.5

. Fortunately this has been replaced by inter-carrier "roaming agreements" and business relationships which allow cellular subscribers to simply use their handset while in another carrier's region without any special pre-arrangements.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... conventional^1.6

. We really don't intend to be snooty in referring to current non-mobile networking technology as "conventional." We have to call it something. Our use of the word "conventional" is merely for clarification purposes-to distinguish non-mobile networks from mobile data networks. We realize that by doing so we risk sounding like the newest/hottest LAN technology sales folks who refer to Ethernet as a "legacy system."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... mobile.^1.7

. This clearly is unacceptable for voice connections, which are routinely made while in transit. However, restricting the movement of cellular voice users would provide the unintended side benefit of increasing roadway safety!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... office.^1.8

. Variations of this scheme have been used by many creative people to further development of their golf games.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... anywhere^1.9

. By definition, if a mobile device can receive data (service) anywhere, that service must be available everywhere. Here, "anywhere" is a geographic concept. Ubiquitous provision of service would thus seem to encourage wireless media for mobile services.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... hosts.^1.10

. While conventional hosts might be relocated over a period of weeks or (more typically) months, a mobile host would relocate over a period of minutes to possibly as short as seconds.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... boundaries.^1.11

. For example, a so-called Class C IP address block is the smallest contiguous block assignment which can be made to an organization. The Class C block defines a three-byte netid and a single-byte hostid, enough for 255 hosts. (All zeros in the hostid field means the network address rather than a host address). If a network administrator only had a dozen hosts to actually assign addresses to, they would only use 12 out of the 255 host addresses available to them. This is not an uncommon situation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ARP^1.12

. The Address Resolution Protocol in the IP world.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... special.^1.13

. One way to implement TAS in a way which is transparent to routers is to have mobile hosts assign their own temporary address, possibly based on the router's (subnet's) netid and their own permanent hostid (which must be globally unique). This is similar to Novell's IPX networking, which greatly reduces the administrative overhead of operating and maintaining a network.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... TAS.^1.14

. The Domain Name System or DNS provides mapping between host names-which rarely change-and host addresses-which change whenever hosts are moved to new subnets in conventional networks.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... uniqueness.^1.15

. Network addresses must be globally unique to prevent data intended for one recipient to be correctly but wrongly forwarded to a different host.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... permanent^1.16

. "Permanent" is a relative term, which usually implies human administration! A configuration can be considered to be "permanent" whenever human intervention is required to change it.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... subnet^1.17

. Actually a protocol such as Address Resolution Protocol (ARP) is used to resolve the hostid to a Layer 2 identifier in use by that host on the subnet. This mechanism is described in [PERL92] and [STEV94].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... balancing)^1.18

. Another strategy for configuring LANs is based on building geography, e.g., all PCs in one area of a floor are connected to the same LAN. With hub-based LAN architectures, a given LAN tap could be connected to any one of the (logical) LANs hosted by the hub.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... end-point^1.19

. We are being purposefully vague here because the host could be a device, an application, or other entity, depending on context.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... authorities^1.20

. The FCC is the governmental body which allocates RF frequencies for various applications and industries in the United States. The World Radio Council (WRC) is responsible for coordinating frequency assignments internationally.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... service^1.21

. Commonly referred-to by "techies" as "Snail Mail." In this case, "service" is a loose terminology.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

....^2.1

. Similar growth and subscription rates are becoming the norm in other countries also. According to the International Telecommunications Union, in late 1995 there were over 55 million cellular telephone subscribers, as compared with 648 million wired telephone lines. Cellular subscriber penetration rates at year-end 1994 for a few representative countries were as follows: Sweden 14.7% , Norway 13.2% , Finland 12.8% , Denmark 9.8% , U.S. 8.8% , Singapore 8.7% , Iceland 8.3% , Hong Kong 7.7% , Kuwait 6.6% , Canada 6.5% .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... data^2.2

. Examples include file transfer, database download, etc., which benefit from committed bandwidth.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... nature^2.3

. Examples include emergency public service, etc., which require guaranteed bandwidth availability.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... users.^2.4

. Things are changing. In recent years sub-minute billing rate plans have been made available and are gaining popularity as a means of attracting customers. Unfortunately, data transmissions rarely need even seconds of dedicated bandwidth.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... channel^2.5

. The term "channel" is inherently ambiguous. We will attempt to always precede the word "channel" with an adjective which clarifies the context-such as "physical" or "logical"-unless the meaning is clear.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... interference.^2.6

. In the world of radio (like much of life), there are no absolute rules, only probabilities! RF signal strength doesn't adhere to strict boundaries, it just sort of peters out. Certain geographic features, such as bodies of water, allow RF transmission to carry over greater distances than is sometimes desired (i.e., beyond cell boundaries). Variable conditions, such as mist, can also greatly affect RF signal propagation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cell.^2.7

. Cells are commonly depicted as being hexagonal in shape. Hexagons barely approximate actual cellular radio coverage areas, which are actually circular under ideal conditions with an omnidirectional transmitter in the center. However, the hexagons are effective for the purpose of tiling a system-wide coverage map without gaps or overlaps. That is why they are used as an industry icon.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cells^2.8

. Subdividing cells by creating new cells is expensive. AMPS cell sites cost approximately $ 1M on average, including support infrastructure, real estate, etc.; the embedded base of 18K or so base stations represents an investment of approximately $ 18 billion.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cells.^2.9

. In terms of RF channel assignments, etc. Unless there is a reason to specify "sector," we'll use the word "cell" to refer to either an omnidirectional cell or a sector.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... populations^2.10

. Early microcells were aimed primarily at filling RF coverage "holes"; current microcells are aimed largely at filling capacity "holes," reflecting the successful penetration of cellular into the marketplace. Small cells are ideal for low power (i.e., portable) mobiles.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... servers.^2.11

. This queueing theoretic result is due to the variances in arrival and service distributions, as discussed in [KLEI75].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... service.^2.12

. This is a common occurrence at airports or large trade shows which is combatted by installation of microcells to help meet the demand for service.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... interference).^2.13

. A larger value of n implies a greater reuse distance, since a larger cluster of cells would share the full set of channels, with each channel used in only one of the cluster's cells.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... mobility.^2.14

. Handoff is a function of radio resource management, which is closely related to mobility management. If a user never moved, there would be no need for mobility management and minimal need for radio resource management.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... BCHO.^2.15

. BCHO handoffs are usually controlled by the MSC, not the base station.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... etc.^2.16

. This is important because a mobile requiring a handoff to a cell without any available channels will be dropped. This type of service disruption is to be avoided at all costs. It is considered better to refuse new calls originating in a cell than to drop calls because there are no channels to handoff to. So service might be denied to new originating calls in order to preserve one or two available channels for handoffs. With DCA this is becoming less of an issue.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... hysteresis^2.17

. Hysteresis is the mechanism which requires a high threshold value to be exceeded before a handoff decision is made. The current channel continues to be used until it has degraded enough that other channels are significantly better for the mobile. In this way unnecessary and undesirable handoffs are avoided. Hysteresis could describe the process of crowning of a new boxing champion; the challenger ("the contendah") must defeat the reigning champion soundly, not just barely, to be declared the new champ.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ratio^2.18

. Pronounced "C-to-I ratio.".

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... signal.^2.19

. C/I ratios relate the desired carrier signal (C) to an interfering signal (S). They are valued in terms of decibels (dB) determined by 10 log (C / I).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... signal.^2.20

. An exception to this is so-called "FM capture," in which the stronger received signal "captures" the receiver and is properly decoded despite the presence of the interfering signal.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... engaged.^2.21

. As we shall see, in time-division digital systems, the mobile can get by with a single transceiver, which rapidly switches between transmitting and receiving modes.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... surfaces.^2.22

. Buildings and mountains do a good job of this.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... level).^2.23

. In reciprocity-based power control, the base station transmits a continuous signal, which includes information about the power level it is using. The mobile measures the strength of the received signal and takes the ratio of that measurement to the indicated transmit power level at the base station. This ratio indicates the "path loss" of a signal between the mobile and the base station and allows the mobile to determine the transmit power level it should use.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... markets.^2.24

. Similar first-generation analog cellular systems include Total Access Communication System (TACS) in the U.K., Italy, Spain, Austria and Ireland; Nordic Mobile Telephone (NMT) in the Scandinavian countries; C-450 in Germany and Portugal; Radiocom 2000 in France; Nippon Telephone and Telegraph (NTT) in Japan; JTACS/NTACS in Japan.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... entrepreneurs^2.25

. Most notable among these was Craig McCaw, founder of McCaw Cellular Communications.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... POPS.^2.26

. This is $ 500 per potential subscriber!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... users.^2.27

. Note that this is illegal.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... markets.^2.28

. The nature of the cellular industry is further fragmented into direct market to market relationships by the fact that many A-side licenses are owned by partnerships of service providers. Often these partners are competitors in different markets. Thus, even within a cellular service provider, it is necessary to functionally separate operations by regions or markets.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... areas,^2.29

. A location area consists of a group of cells (or clusters), which, conveys a page for a mobile station. Paging is done on the basis of location areas; presumably one of the location areas is where the mobile station is currently situated.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... CSMA^2.30

. Carrier Sense Multiple Access is a common MAC protocol, used in Ethernet LANs. It is described in Chapter 0.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (PBX)^2.31

. A PBX is a small typically privately-owned switch, which provides many of the advanced voice features available in businesses. It interfaces to public telephone switches via special protocols.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... message^2.32

. The "blank and burst" refers to the blanking of the audio (voice) carriage in favor of a brief burst of data, including the modem training, between the subscriber's handset and the cellular network.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... area).^2.33

. The trade-off of a gradually expanding paging "radius" is increased ringing delay experienced by the originator of the call attempt. If it has been a while since the mobile last registered, the last location area used by the mobile is probably obsolete.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ADPCM^2.34

. Analog to Digital Pulse Code Modulation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 1990's.^2.35

. After a while, continually subdividing cells into sectors and microcells becomes prohibitively expensive and inefficient. The more cells, the more cell handoff signalling and decision-making required, etc.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... boundaries.^2.36

. Prior to GSM each nation had its own analog system, which was mostly incompatible with the systems of neighboring countries. No cross-border roaming was possible with these analog systems.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... conditions.^2.37

. Because each frequency is impacted by environmental conditions, including interference, differently.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... communications,^2.38

. The story is that early GSM industry participants in the UK rephrased GSM to mean "God Send Mobiles." Then, as the mobiles became increasingly available, GSM meant "Good Sales and Marketing." Another example of an acronym taking on a life of its own.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (PCN)^2.39

. PCN is the European equivalent to what is called PCS in North America. More on PCS later.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... handovers^2.40

. GSM calls handoffs "handovers."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... GMSK^2.41

. Gaussian Minimum Shift Keying.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... interleaving^2.42

. Interleaving is a shuffling of the bits in a transmitted packet which distributes and randomizes those bits impacted by noise and interference. This has the net effect of turning an error burst into random bit errors, which are much easier to correct via standard techniques.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... rate^2.43

. Digital voice is data!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... [FALC95].^2.44

. For this comparison AMPS is assumed to require one 30-kHz band per user with an 18 dB C/I ratio and a frequency reuse factor of 7 (cells) in a 3-sectored arrangement.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... PCMCIA^2.45

. Personal Computer Memory Card Industry Association.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 6.3^2.46

. The calculation: [3 users / (25 kHz * 4 cells)] / [1 user / (30 kHz * 7 cells)] = 6.3.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... imminent.^2.47

. Strictly speaking, the TDMA standard is a combination of FDMA (frequency division multiple access) and TDMA. Likewise, the CDMA standard is a combination of FDMA and CDMA.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... TIA^2.48

. The Telecommunications Industry Association is the North American standards body which oversees the cellular industry.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... standard.^2.49

. Given the mixture of technical theology and ego involved, it is difficult to see any of the service providers switching their technical allegiance. Thus, nationwide digital service will likely depend on triple-mode cellular handsets, with CDMA, TDMA and the old standby, AMPS.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... series^2.50

. These standards are numbered IS-130, IS-135, IS-136.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... D-AMPS.^2.51

. Supporters of CDMA would probably challenge this informal naming convention.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... call.^2.52

. A TDMA frame is 40 msec or 972 2-bit symbols long and consists of six timeslots. Each full-rate TDMA channel consists of two of these six timeslots.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... N-AMPS^2.53

. Motorola's narrowband AMPS system, which is not a standard.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... II^2.54

. Frequency hopping spread spectrum patent number 2,292,387 belongs to none other than Hedy Lamarr, a screen siren for MGM in the 1940's. Who says you need an engineering degree?

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... required.^2.55

. In fact, one could argue that the lack of commercial CDMA deployment at year-end 1995-following numerous delays-reflects the degree of difficulty in mapping between these assumptions and the vagaries of real-world cellular environments.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... U.S.^2.56

. Motivated largely by huge budget deficits, Congress authorized the FCC to use an auction process for allocating PCS licenses in the 1993 Budget Reconciliation Act. This auction process replaced the earlier lottery for AMPS licenses, which resulted in widespread fraud (by promoters on an unsuspecting public) and abuse (stuffed ballot boxes). The A- and B-block auctions lasted 111 rounds and raised over $ 7 billion for the U.S. Treasury, averaging $ 15.50 per POPS; the most expensive market was Chicago at over $ 30 per POPS. The C-block auctions were held later and, despite being targeted for small businesses, raised over $ 10.2 billion.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (licenses).^2.57

. The FCC prohibits a single entity from operating both a cellular service and a PCS service in the same market.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... systems.^2.58

. PCS can be expected to suffer from a 10 dB propagation penalty because of the higher frequencies. This amounts to a coverage range of about one-half of cellular and a coverage area of about one-quarter of cellular.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... IPv6^3.1

. Formerly known as Internet Protocol next generation or IPng [BRAD96].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Inc.^3.2

. Now known as AT& T Wireless Services, Inc.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... technology.^3.3

. Many of the original system concepts, such as channel hopping, originated at Novatel; rights to this technology were subsequently transferred to IBM.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... channel-hopping^3.4

. In frequency hopping, the logical data channel "hops" to an idle RF channel upon either a timer expiry or the incidence of a voice call coming up on the former RF channel. This is different from the "frequency hopping" employed by some spread spectrum systems.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... services^3.5

. The earliest specifications for CDPD included an entity called the Mobile Data Gateway, a precursor to the current Mobile Data Intermediate System. Despite its name, the actual network gateway services were to be located in entities called Network Interface Modules, one of which was to be defined for each of IP, SNA, etc.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... CDPD^3.6

. At the Comdex trade show that fall in Las Vegas, the seven sponsoring cellular service providers demonstrated the CDPD prototype in a common booth. A second more-detailed release of the specification for the telephony-based system architecture was scheduled for publication immediately following the show. However, one of the cellular carriers abstained in the vote to approve the publication. A cautious voting process, in which an abstention could block action, arguably may have preserved the credibility of the cellular service providers as potential mobile data services providers.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Parts^3.7

. The CDPD System Specification and Implementor Guidelines are divided into "Parts," which discuss specific system aspects or protocols.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Guidelines^3.8

. Unfortunately, some developers of CDPD equipment and software have mistakenly interpreted the CDPD Implementor Guidelines as just guidelines, which was not the intention of the specification team. A more formal standardization process has been established in the CDPD Forum to clarify any such misperceptions.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... technologies.^3.9

. This misperception has been fed to a large extent by misleading comments from some purveyors of alternative technological solutions for the airlink. It seems that as technology advances, there are increasing opportunities for apples-to-oranges comparisons by "technologists" with agendas. It is our hope that this book will clarify some of the misperceptions about mobile systems in general and CDPD in particular that have been propagated over the past few years.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... WANs.^3.10

. Mobility management and radio resource management provide functions of both network services and network support services.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... bear.^3.11

. To a large extent we have always considered the RF technology employed in CDPD to be a "temporary" solution. Once again protocol layering supports the evolution to more sophisticated RF technologies as the costs of these technologies decrease.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... percent.^3.12

. With the more than forty percent compound growth rate in the North American cellular subscriber population during the early 1990's, maintaining a busy hour performance level of two percent blocking has proven to be a significant challenge. Most CDPD service providers have resorted to dedicating RF channels for CDPD service. This short-term sacrifice is offset to some degree by the growing capacity gain provided by digital voice services.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cloud"^3.13

. No presentation of network technology is complete without at least one "cloud" diagram, so here it is!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... CLNP.^3.14

. Routing of reverse channel IP packets from a visiting mobile back to a host in its home system could be directly routed using standard IP-based protocols, in which case the interface between the two service providers would function as an E-interface. But forward-direction packets would be redirected from the home to the serving via CLNP encapsulation, as discussed later.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... providers.^3.15

. As mentioned in Chapter 2, a typical cell site costs approximately one million dollars or more for initial construction and introduction into service. With an industry total numbering over 18 thousand sites, more than $ 18 billion have been spent thus far on cellular infrastructure.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... System^3.16

. With the increasing miniaturization of the CDPD modems, it is unclear if removable SIM cards will ever be produced. Indeed, for the removable SIM concept to gain acceptance, manufacturers need to produce CDPD devices with the proper interfaces to accept SIM cards. One possible use for SIM card devices is in fleet operations. In this case, it may be beneficial to deploy mobile units with SIM slots into each fleet vehicle, while the drivers are assigned individual SIM cards. This way, each driver is uniquely identified by his or her network address.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... element.^3.17

. Strictly speaking, in terms of user data transmission, the MDBS is a link layer relay and is not a part of the network layer architecture. However, it serves an important role in this wireless mobile data network and deserves attention here and in the CDPD System Specification.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... router^3.18

. When we began work on the CDPD system specifications, it was clear that we needed to use concise terminology. We were not interested in unnecessarily inventing new terms, so we used existing terminology as much as possible. We chose the ISO terminology as a base, not because we were "ISO-bigots" but because we were not as familiar with them, and thus less likely to be burdened with unspoken and differing interpretations which could cause confusion. Besides, OSI terminology sounds much more sophisticated.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... BGP-4^3.19

. BGP-4 (Border Gateway Protocol, version 4), in support of classless interdomain routing (CIDR), is essential for efficient allocation of the somewhat limited IP address space to M-ESs, due to the "permanence" of these address assignments. In the future, one could expect CDPD support for the IPv6 (formerly IPng) protocol.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... System^3.20

. In the spirit of political correctness, some of us wanted to call the F-ES a "Mobility-Challenged End System." Fortunately better judgement prevailed on the CDPD specification team and "F-ES" was established.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... environment.^3.21

. The particular example of concern to the Mobile IP task force was the scenario in which two travelers from the US were in Amsterdam and wanted to share data. The round-trip path for the data packets would go from Amsterdam to the US and back to Amsterdam. Since CDPD was developed under the auspices of North American cellular carriers, this was considered to be an extreme case and not representative of the mobility CDPD was originally intended to address. Quite frankly, we look forward with great anticipation to the day that this becomes a significant issue.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... system,^3.22

. In TDMA the mobile assists the system by evaluating RF channel quality and sharing this information with the system. This seems to be the best way to handle RRM, since it is the mobile which is best situated to determine the best channel to use.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Language^4.1

. The American Heritage Dictionary of the English Language, Third Edition copyright ©1992 by Houghton Mifflin Company.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... sector.^4.2

. A CDPD cell might or might not share boundaries with an AMPS cell or sector. Depending on the frequency plan in use, a CDPD service provider might instead elect to utilize a "booming" CDPD frequency plan, in which a CDPD cell covers multiple AMPS cells or sectors. This has been done by many CDPD service providers to reduce deployment costs.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... MDBS^4.3

. To be precise, in a sectored cell, a single common MDBS might control the activities in each of the sectors, which are treated as independent CDPD cells. This is how vendors of MDBS equipment have implemented their solutions, in order to reduce the costs per cell site for a CDPD service provider.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (GMSK)^4.4

. Both GMSK and DSMA are defined for AMPS-channel implementations of CDPD. Alternative airlinks are under development in the TDMA Forum and the CDMA Forum, but as of our publication deadline have not been formally released. These alternative airlinks are likely to utilize other physical and MAC layers than GMSK and DSMA.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... MD-IS^4.5

. Early versions of the CDPD System Specification defined a network entity called the "Mobile Data Intermediate System". With CDPD System Specifications preliminary release 0.8, we introduced the concept of a serving area entity and a home area entity. Unfortunately, there was concern that confusion may result from the appearance of a "new" network component. Therefore, instead of creating yet another acronym, we opted for the notion of a "serving MD-IS" and a "home MD-IS". To our dismay, this has also caused some confusion. Some readers misinterpreted the names to mean that if a M-ES is in its "home" area, it would only interact with the "home MD-IS". This is not so. A perhaps better definition of function involves the mobile home function or MHF and mobile serving function or MSF; these definitions provide clarity in the case of a single MD-IS supporting both homed and visiting mobiles.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... NEI.^4.6

. An M-ES might have more than one NEI assigned to it, especially if it belongs to one or more multicast groups. Each NEI must be separately registered, authenticated and tracked by the system.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... control.^4.7

. If the serving MD-IS and the home MD-IS belong service providers who have no business arrangement, the serving MD-IS may not bother sending the RDR message. In this case, the serving MD-IS could possibly return a negative result code to the M-ES in the ISC message. Usually, however, the serving MD-IS would be expected to try to provide service and the home MD-IS would make the go/no-go decision.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... value^4.8

. Given the fixed size of the data field, it "wraps around" to zero after reaching the maximum value.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... device^4.9

. Mobile to mobile communications is also possible and is supported in the same manner as from an external host. In other words, network data packets from mobile hosts do not receive special routing support. All network data packets destined for a mobile device must be first redirected through the designated home MD-IS for that M-ES.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... astunneling^4.10

. As we discussed in Chapter 1, there are basically two methods to accomplish the packet redirection needed to support mobility: encapsulation and simply changing the packet's destination address. The CDPD specification team elected to go with encapsulation because we felt that it was a "cleaner" design. Changing the packet destination address would require insertion of the M-ES's NEI somewhere in the packet to enable the serving MD-IS to correctly handle the packet.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... overhead^4.11

. Except for retransmissions of data traffic lost due to the movement between cells.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... subdomain^4.12

. The M-ES recognizes movement across cell boundaries and routing area subdomain boundaries through interpretation of Radio Resource Management broadcast data and messages. Radio Resource Management is discussed in the next chapter.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... link^4.13

. Data link establishment also involves exchange of encryption keys and initiation of data encryption. These are described in the chapter 6.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... addresses^4.14

. There is no theoretical limit on the number of NEIs used by an M-ES. But practical limitations do apply, usually as a result of the IP protocol stack implementations that have been ported to M-ESs.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (TEI)^4.15

. The TEI is described in Chapter 5.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... encryption^4.16

. The security goals and mechanisms within the CDPD system are detailed in chapter 6.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... M-ES^4.17

. To be exact, the home MD-IS and the serving MD-IS are verifying their ability to reach the NEI on the M-ES.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... NEI^4.18

. An active NEI is an address that is currently registered on the network.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... credentials^4.19

. Release 1.0 of the CDPD System Specification failed to include the ability to validate individual multicast group members. We initially felt that since multicast address are not allowed to initiate data packets, there is really no security risk. However, a mobile could demand and cause the transmission of forward channel multicast frames. If the mobile device is not authenticated, it is possible to cause unnecessary forward channel traffic. Moreover, without group member authentication, it is not possible for the service provider to repudiate claims that the forward traffic is invalid. Now, if the service provider cannot accurately account for the traffic and accurately control the channel usage, they can't run the network. Thus the need for individual group member authentication.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... sight.^4.20

. In the end, IPv6 was established in July 1994.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... hops.^4.21

. If you don't know what we mean, try using the traceroute utility!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Parts^5.1

. Parts 400, 401, 402, 403, 404, 405, 406, 408 and 409 define the airlink interface.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... conversation!^5.2

. Unless of course it is a gathering at the CDPD Forum membership meeting!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... transit.^5.3

. You can experience this on overseas long distance calls that are routed through communications satellites. The quarter second round trip delay to the satellite can make conversations quite awkward.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ^5.4

. An excellent presentation of Aloha and slotted Aloha may be found in [KLEI76].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... retransmissions^5.5

. Real networks typically will not allow infinite retransmissions. Finite retransmissions along with disconnection of the link will result in performance different from the theory.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... transmission.^5.6

. This ignorance of channel status results in an enlarged collision window, which in turn decreases the probability of successful transmission, causing the eventual breakdown of the shared channel.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... CSMA/CD^5.7

. The Decode Status flag actually provides a bit more data since any error that results in decode failure by the MDBS is fed back to the mobile units via this flag.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (DSMA)^5.8

. Actually, the full description of the airlink MAC protocol in CDPD is "slotted non-persistent DSMA with collision detect." Yikes!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... errors^5.9

. Although the Reed-Solomon (63,47) encoding could be used to correct up to 8 symbol errors, doing so would raise the probability of undetected symbol error to approximately 1.2 x 10^-5. It was the desire to avoid undetected symbol errors which drove our adoption of Reed-Solomon coding rather than convolutional coding in CDPD.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... block^5.10

. Rapid collision detection reduces the size of the collision window of DSMA (in CDPD) to 2 microslots (the parts of the block which lie between control flags). Students of contention protocols will recognze that it is the minimization of the collision window which is most responsible for increasing the efficiency of a contention protocol.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... flag.^5.11

. It is this synchrony between the state of the reverse channel and the broadcast control flag which reduces the amount of "busy hang" time. Busy hang is the state which exists when the reverse channel is actually idle but a mobile awaiting transmission believes it to be busy because the flags indicate it to be busy. The size of the busy hang interval limits the efficiency of the shared medium of the reverse channel.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... microslots^5.12

. A microslot in CDPD is equal to the transmission time of 60 bits at the 19,200 bps.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (LAPD)^5.13

. Link Access Protocol on the D-channel (LAPD) is an ISDN-standard protocol, which defines the link operation for multiple "dumb" terminals sharing a common physical link to a "master" ISDN device. This standard was deemed ideal for the purposes of the CDPD channels with the following extensions: longer and variable-length IDs for the mobiles, Selective REJect capability for more efficient use of the airlink, no parity word because of the strong Reed-Solomon encoding, sleep mode for mobile's battery life conservation, multicast capability for groups of mobiles.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cell.^5.14

. The "system" referred to is the MD-IS, or more precisely, the Mobile Serving Function of the MD-IS. The range of control for an MD-IS is a mobility area.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 1.1^5.15

. Enhancements to MDLP included the replacement of the XID parameter negotiation process with the negotiation taking place during the TEI request/assignment process, additional timers to enhance performance, and clarifications of procedures for handling various error recovery situations.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... mechanism.^5.16

. We not only introduced the selective reject mechanism, we removed the normal reject mechanism. We had long and difficult discussions about whether we should retain the normal reject mechanism to reduce the chance of destabilizing existing protocol software base (we anticipated most developers would start from a LAPD implementation). In the end, we felt that if the normal reject mechanism was retained, there may be little incentive for the developers to use add the selective reject mechanism, then our intent of a more efficient error recovery mechanism would be foiled.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... frame^5.17

. To be precise, the probability of a block being decoded with an undetected error and passed to the LLC layer is 1.2 x 10^-5 (assuming 8 symbol correction at cell edge of 17 dB C/I) and 2.75 x 10^-8 (assuming 7 symbol correction at cell edge of 17 dB C/I). The CDPD specification recommends 7 symbol correction.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... protocol^5.18

. Unfortunately, in [CDPD93] and [CDPD95], SNDCF is called a "protocol" with the obvious acronym "SNDCP". We're splitting hairs here... 'Nuf said!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... PDU^5.19

. This assumes the subsets, mandatory functions, and optional functions specified by [IGOSS-1] and [NIST-500-206] are in effect. (i.e. Inactive subset is not used, Non-segmenting subset is not used, use/non-use of checksum is configurable, and option fields are permitted)

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... message^5.20

. A UNCOMPRESSED CLNP message is different from an unaltered CLNP message.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... line^5.21

. In the real world of RF shadow effects and varying reflectors around the mobile device, measurements will fluctuate between 5 dB to 8 dB. So the line of equal power between the two cells "jumps" around a long term average.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... size^5.22

. It seems counter intuitive that a negative value favors the current cell. Unfortunately that is how the equation for the algorithm was documented.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cell^5.23

. The comparison is carried out according to the following formula:
If RSSI_curr + RSSI Hysteresis > RSSI_RefCh - ERP Delta_R-C + Bias _Adj-Curr
then stay in current cell, otherwise the current cell is not the "best".
The use of the RSSI Hysteresis value is discussed in a later section.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... blocking,^5.24

. Given the explosive growth of cellular subscribership, it is not always possible to maintain such a low blocking factor. Cellular carriers are continually adding infrastructure, dividing cells, and other technologies to keep up with the demand.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... intelligent^5.25

. AI experts within the cellular community started salivating.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Sniffer^5.26

. What a name!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... immediately^5.27

. Within 40 milliseconds.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... calls.^5.28

. Instead of a sniffer, the MDBS is also a stinker.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... this^5.29

. We have often heard that a layered communications architecture design is important. However, the development of CS CDPD is concrete proof that such design techniques are valid and useful.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... function^5.30

. To be exact, the system makes use of the cellular telephone system's frequency based demand assigned multiple access scheme.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (CSCCP^5.31

. It is usually of concern when we have to resort to acronyms of acronyms!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... CDPD^5.32

. Isn't layering wonderful!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cracker^6.1

. We share the sentiment expressed in [KAUF95] that "hacker" should be reserved for master programmers who are honest. "Cracker" is a derivative from "safe cracking," romanticized in print and reel, which is unfortunate.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ATM^6.2

. Automated teller machine, not a much-hyped link layer protocol.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... "Two-Way."^6.3

. Also called bilateral authentication.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... PDUs^6.4

. Subnetwork data protocol data units.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... identifiers^6.5

. NEIs are most commonly IP addresses.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... M-ES^6.6

. To be correct, the ESQ message is directed to an active NEI within the M-ES. Authentication credentials are associated with the NEI and not with the M-ES device.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... months^6.7

. Some of the early networks were configured with this high authentication credentials update rate. Better understanding of the issues through this experience has corrected the problematic network configurations.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... link,^6.8

. More precisely, key exchange is done prior to establishing the multi-frame mode in MDLP.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... We^6.9

. "We" in this context refers to the CDPD specification team, not just us!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... system.^6.10

. Of course we can't prevent sales folks and marketers from saying what they will!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... vulnerability.^6.11

. One of the purposes of enumerating the non-goals for CDPD security was to acknowledge potential vulnerabilities.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cloning.^6.12

. Although doing this would require decoding of CLNP datagrams - a relatively minor security mechanism!

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... provisioning^6.13

. Provisioning refers to the process of activating a new user on the network.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... following:^7.1

. There is some overlap between these two lists. This is because support functions tend to be general purpose in nature and address multiple support needs.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... management^7.2

. Network management is called "systems" management in OSI terminology, which does not distinguish between computers and networks. Throughout this chapter (and quite possibly the entire book as well), when we refer to "management" we mean management of networks, not people.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... system^7.3

. The "management system" is the application running on one or more computers, which controls the systems management activities. Of course, the management system must itself be managed.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... agent^7.4

. The agent is the application which monitors and controls the state and operation of a managed object. It facilitates the management functions for that object under the control of the manager.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... manager^7.5

. The manager is the application running on one or more computers which monitors and controls overall system operation. Its control is usually exercised via agents, which are more closely attuned to individual objects.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Internet.^7.6

. An example of the proverbial "security through obscurity."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... CDPD.^7.7

. In bad RF conditions or in a contention situation, multiple attempts may be required to get data packets across the airlink. Even though these multiple attempts are not counted (more than once), they could require enough time that Layer 4 time-outs expire at the mobile or the correspondent host.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... customers^7.8

. The CDPD System Specification distinguishes between a customer and a subscriber in the same way that the cellular industry does. A customer is the entity with which a CDPD service provider has a business relationship. A subscriber is the individual actually receiving service. The customer pays the bills on behalf of the subscriber; a typical example would be a large corporation which would be the customer, with many of its employees receiving service as subscribers.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... areas.^7.9

. Although monthly exchange of HASs between CDPD service providers fits neatly into existing cellular billing mechanisms and practices, the CDPD specifications support much more frequent exchange of this information. Frequent exchange of detailed usage accounting information would assist in preventing fraudulent use of CDPD services by "bad guys."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... names^7.10

. Actually having less user-friendly names adds some measure of security; for this reason it is desirable to not have every network resource name contained in a user application-accessible directory.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... it.^8.1

. Remember, submission is from client (at the mobile) to the server; delivery is from the server to the client.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Area^8.2

. In release 1.1 of the CDPD System Specification, the word "Area" was added to the title of the service. This is in recognition that the accuracy of such an approach is limited to a cell coverage area. The intent was to avoid confusion with systems that provide "pinpoint" accuracy.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... report^8.3

. Event reports are also generated for registration and deregistration events.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... degree)^8.4

. Remember, these are all approximate since RF is a world of probabilities.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... application?^8.5

. The likely answer is "All of the above." There wasn't enough time to address them all.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... FCC^9.1

. These frequencies were originally in Part 18 of the FCC regulations.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... standard^9.2

. IEEE 802.11 also has provisions supporting infrared media.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... simulcasting^9.3

. Simulcast means that the same message is transmitted simultaneously from a number of locations in the network.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... equipment^9.4

. For example, Canadian paging systems utilize 1200 baud rates vs. the U.S.'s 2400 baud. For this reason many pager models accommodate both baud rates.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... States.^9.5

. Economics and Management Consultants International (EMCI) estimate.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... degree.^9.6

. Despite its receive-only mode of operation, the perpetual duty cycle of a one-way pager results in a significant power drain over time. Thus, the need for an effective sleep mode.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... bands.^9.7

. NPCS frequencies include 900-901, 930-931 and 940-941 MHz bands. Only two of these three frequency bands have been auctioned and assigned. The third is being held in reserve for later assignment.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (pACT)^9.8

. This is pronounced "pact," not "p-act."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... definitions.^9.9

. This extensibility of the CDPD specification is a benefit of an open, layered system definition. Network elements and protocols in the pACT technical specification are the same as in CDPD, but with the leading "Mobility" being replaced by "Personal," e.g., "PD-IS" instead of "MD-IS."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... non-trunked^9.10

. A trunked system is one with shared RF channels among disparate users.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... SMR^9.11

. This is pronounced "smur."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... station.^9.12

. Like the party lines commonplace in bygone days, the multiple users of the channel determine that the call is their's by the ringing pattern or by the identifier called out. Remember "Car 54, where are you?"

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... year.^9.13

. American Mobile Telecommunications Association and Economic and Management Consultants International, Inc., estimate.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (ESMR^9.14

. This is pronounced "e-smur."

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (MIRS).^9.15

. This formerly proprietary technology was renamed iDEN or integrated Dispatch Enhanced Network in 1995. iDEN may now be licensed from Motorola. MIRS was originally developed for NexTel (then known as FleetCall). Motorola subsequently traded its SMR channels (then amounting to 30% of the channels available in the U.S.) to NexTel in exchange for an equity interest in the company.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... mid-1995.^9.16

. The slow acceptance of these services has spurred RAM and Ardis to interconnect their services with those of cellular and CDPD service providers.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... band.^9.17

. RAM assembled its nationwide spectrum by purchasing and consolidating SMRs.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... hackers.^9.18

. RAM Mobile white paper [RAMM95].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... required.^9.19

. This depends on the so-called link power budget and aperture required (to differentiate between two satellites with small angular separation transmitting on a common frequency).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... constellation.^9.20

. The joke inside Motorola was that it was a good thing that 76 satellites were insufficient. The element whose atomic number is 76 is Lead, which somehow lacks the marketing cachet of Iridium. Likewise, the element whose atomic number is equal to the currently-planned number of satellites-66-is dysprosium, which is probably a name better suited for things other than communications services.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... network.^10.1

. Specific external methods of acquiring a local IP address for use as a co-located care-of address are beyond the scope of this book.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.