7.3 PBX Components

---

Aside from the line and trunk interfaces, three major elements comprise the typical PBX: processor, memory, and switching matrix. Together, these elements provide all of the intelligence necessary to place calls anywhere on the public or private network without the need for human intervention.

7.3.1 Central Processor

The central processor is responsible for controlling the various operations of the PBX. This includes monitoring all lines and trunks that provide connectivity, establishing line-to-line and line-to-trunk paths through the switching matrix, and tearing down connections upon call completion. The processor even controls such optional capabilities as voice mail and the recording of billing and traffic information. Because the processor is programmable, features and services can be added or changed at a management terminal.

Many PBXs may be optioned for two or more processors to increase the reliability of the system; if one fails, another takes over. Programs and configuration information are automatically downloaded from the main processor to the standby processor to ensure uninterrupted service.

7.3.2 Memory

The central processor uses the memory element to assist in the implementation of the advanced functions of the PBX, many of which are defined and implemented in software, instead of hardware. There are two types of memory: nonvolatile and volatile. The former is fixed, whereas the latter may be changed as needed. Nonvolatile memory contains the operating instructions and stores system configuration information. Volatile memory, also known as random-access memory (RAM), is used for temporary storage of frequently used programs or for workspace.

In the event that a system failure was to destroy the content of the nonvolatile memory, a reserve program, also stored in nonvolatile memory, would begin operation automatically. (Some systems, however, still require that a spare program be loaded manually via disk or tape cartridge after a catastrophic failure.) As the term implies, nonvolatile memory can withstand a power outage while, in the process, retaining its contents. Information stored in nonvolatile electronically erasable read-only memory (EEROM) is automatically dumped into RAM for access by the redundant processor. Although EEROM is nonvolatile, it is changeable.

7.3.3 Matrix

The switching matrix, controlled by the processor, interconnects lines and trunks. ^[1] This may be accomplished through space-division switching or time-division switching. Space-division switching originated in the analog environment. A space-division switch sets up signal paths that are physically separate from one another, or divided by space. Each connection establishes a physical point-to-point circuit through the switch that is dedicated entirely to the transfer of signals between the two endpoints. The basic building block of the space-division switch is a metallic cross-point (relay contact) or semiconductor gate that can be enabled and disabled by the processor or control unit. Thus, physical interconnection is achieved between any two lines by enabling the appropriate cross-point.

Although the single-stage space-division switch is virtually non-blocking, ^[2] it has several limitations, the most serious of which is the number of cross-points that are required as the number of input lines and output lines grows (see Figure 7.2). This is not only costly but results in ever greater inefficiencies in the utilization of available cross-points. Growth notwithstanding, the loss of a cross-point prevents the interconnection of specific stations.

Figure 7.2: Architecture of a single-stage space-division switch. Because the single-stage space division switch always has a path available for the interconnection of inputs to outputs, it may be considered nonblocking. This scheme is acceptable only for small configurations. As the number of cross-points increases to accommodate system growth, the matrix becomes less efficient.

These limitations are mitigated through the use of multistage cross-point matrices-(see Figure 7.3). Although requiring a more complex control scheme, the use of multiple stages reduces the number of cross-points, while increasing their utilization. Also, because more than one path through the network exists to connect two endpoints, reliability is increased as well.

Figure 7.3: The architecture of a three-stage space-division switch. In this type of switch, the probability of blocking is substantially reduced. Although blocking can be further reduced by adding more switching matrices via multiple stages, blocking cannot be eliminated entirely.

Whereas space-division switching originated in the analog environment, time-division switching is a digital technology. It is the preferred switching technique employed in the latest generation of PBXs. With digital technology, voice signals are sampled at a rate of 8,000 times per second and encoded for digital transmission using PCM. The resulting 64-Kbps channel (8,000 samples by 8 bits) corresponds to a single-voice conversation, which is interleaved (multiplexed) with other input channels for simultaneous transmission through the matrix. The benefits of digital-time division extend beyond the matrix to the transmission link itself.

A digitized voice signal is not only easier to switch, it is easier to store, recognize, synthesize, multiplex, concentrate, and integrate with other digital streams; and, once digitized, these functions can be performed without increasing the impairments that typically disrupt transmissions over analog facilities. Furthermore, using PCM to encode an analog signal into digital form makes the signal compatible with the multiplexing formats used by telephone companies for transmissions through the public network.

For private PBX networks, there are add-in cards that implement various voice compression schemes, such as adaptive differential pulse code modulation (ADPCM). This standard algorithm reduces the 8-bit words under PCM to 4-bit words, thus doubling the number of voice channels available over a digital line. Instead of carrying voice at 64 Kbps under PCM, ADPCM allows voice to be carried very reliably at 32 or 16 Kbps to increase the bandwidth efficiency and eliminate the need for extra trunks between PBXs at each corporate location.

^[1]Although the terms “line” and “trunk” are often used interchangeably, there is an important difference. A line refers to the link between each station (telephone set) and the switch, whereas a trunk refers to the link between switches. The term “tie line,” then, is really a misnomer. It should really be called a “tie trunk.”

^[2]"Blocking” refers to the inaccessibility of the switch due to the unavailability of cross-points, which establish the connections between various endpoints. In theory, all switches can experience blocking no matter how they are designed. As a practical matter, however, some switch designs are less prone to blocking than others. Blocking can also be the result of having fewer trunks than lines, which is a normal condition. The assumption behind high line-to-trunk ratios is that not all users will try to access the switch at once. But if 21 lines out of 200 try to access 20 trunks, the twenty-first line will be blocked.

---