![]()
Chapter 1
Network Evolutions, Applications and Their Synchronization Requirements
1.1. Introduction
The subject matter contained within this book relates to synchronization in telecommunication networks. Synchronization is widely studied and plays a significant role in how the telecommunication network allows us to communicate. As a topic, the rationale behind some aspects of synchronization often appears to be shrouded behind a veil of secrecy, understood only by a few practitioners.
Synchronization is a widely used term in language that has the connotation of alignment in time. As a term, databases can be synchronized, watches are said to be synchronized and networks can be synchronized. Within this book, we limit the term synchronization to pertain to the ability to transfer either time, frequency or phase from one system, or clock, to another system. On a network-wide scale, network synchronization means that all network elements may be configured to share a common frequency, phase or time relationship.
In most telecommunication networks, the need to distribute the information necessary to allow all network elements to be “synchronized” results in the creation of a dedicated synchronization network. This network provides a path from the network clock (e.g. Primary Reference Clock (PRC)) to the individual network elements requiring timing, and may involve the use of other dedicated clocks (e.g. a Synchronization Supply Unit (SSU)). In many cases, the service transport technology has been designed to provide the capability to carry network timing via dedicated synchronization interfaces. For example, Synchronous Digital Hierarchy (SDH) has special input and output timing specific interfaces. This results in a tight linkage between the synchronization network and the transport network.
The requirements for the synchronization network are based on the needs of the services carried over the network. Services offered over the telecommunication network are rapidly changing and evolving. In a sense, the requirements may be evolving, but the capabilities of the network may be set by a specific technology or service.
As the network evolves, it may be found that the synchronization provided by the network is not sufficient for new services. To understand the current capabilities of the synchronization network, it is useful to understand how the telecoms network has evolved to the modern data communication network it is today.
Telecommunications began with telegraphy, which was simply a very rudimentary, low bit rate data service. The ability to telecommunicate by natural voice was, once demonstrated, something highly desired by the cognoscenti. This then spurred the rapid development of technology, first based on manual analog connections, then later based on user-initiated connections in automatic exchange switches. The rapid increase in popularity of telephony and increase in traffic drove the development of transmission systems capable of carrying multiple voice channels simultaneously and miraculously on a single wire. As technology continued to evolve, digital technology was gradually introduced in the mid-20th Century with the development of Pulse Code Modulation (PCM) where the analog voice channel was sampled and converted to a digital bit stream. This technology allowed replacement of analog carrier systems, greatly reducing the cost of sending information by wire. During the 1970s, the first all digital voice switches were developed and represented a quantum leap in technology. Service quality increased dramatically, as the effect of analog impairments was drastically reduced. The technologies used to offer data services were also evolving during this time. Data transmission relied on the use of the analog voice channel to transmit data. Initial data rates in the order of 300 bits/s very gradually increased as modulation techniques improved. At that apex of technology, analog voice band modems were capable of providing up to 56 kbit/s over a nominal 4 kHz channel. Digital data, the carriage of data directly without modulation, became possible with the Integrated Services Digital Network (ISDN) that resulted from a combination of digital voice switches, coupled with digital PCM transmission systems. Many volumes have been written on these technologies.
During the 1970s, the first packet-switched data networks were being placed into service, again leading to dramatic technical development in network technology, notably Asynchronous Transfer Mode (ATM), which transmits and switches packet of fixed length (cells). Packet or cell switching was understood to provide advantages due to the statistical nature of data transmission, although these were primarily for computer-to-computer or terminal-to-computer communications. ATM standards, however, added aspects that allowed its use as a general purpose transport layer, including support for constant bit rate (CBR) services.
The ability for users to access the network has changed dramatically. The traditional telephone channel was restricted to a 4 kHz analog channel. The concept of wider access bandwidths had always been an objective. Digital Subscriber Lines (DSL) was an advanced modem technology that boosted access rates to the low megabit per second range at relatively low cost. Communications were between the subscriber and special equipment (digital subscriber line access modules). Similarly, cable modem technology was also evolving to support a bi-directional channel over the coaxial infrastructure that was common for television cable systems. These wireline access technologies resulted in the transformation of the network from a voice-based network to a network that now predominantly carries data.
However, perhaps the most dramatic swing was toward the use of wireless technology as a network access infrastructure. During the early 1980s, cellular telephone systems provided mobile access to the telecommunication network. Initial systems were analog systems, with voice as the primary service. Later generations of technology migrated to digital transmission and included data communication channels. Following the example of wireline telephony, wireless telephony access grew to the point where the majority of traffic is data, rather than the initially intended voice telephony.
1.2. Evolution from plesiochronous digital hierarchy to optical transport networks
1.2.1. Plesiochronous digital hierarchy and public switch telephone networks
Synchronization became a very important aspect of telecom networks with the introduction of the plesiochronous digital hierarchy (PDH) in transport networks and digital switches in public switch telephone networks (PSTN).
The PDH hierarchy is based on two basic rates, 2.048 Mbit/s (E1) for the European networks based on the European Telecommunications Standards Institute (ETSI) standards and 1.544 Mbit/s (DS1) for the North American networks based on the American National Standard Institute (ANSI) standards. They could be multiplexed into the higher rates defined by the PDH hierarchy, that is 8.448, 34.368 and 139.264 Mbit/s for ETSI and 6.312 and 44.736 Mbit/s for ANSI. It is common to call the multiplexed input signal a “tributary” and the resultant signal an “aggregate”.
PDH has two main advantages. First, it is able to transport timing since the multiplex principle defined in the PDH hierarchies is based on a 1 bit justification process, allowing appropriate jitter and wander performance on the egress 2.048 and 1.544 Mbit/s to be compliant with the synchronization interface specification defined in International Telecommunication Union Telecommunication Standardization Sector (ITU-T) Recommendations G.823 [G.823] and G.824 [G.824]. The second advantage of PDH is that it does not require any synchronization for itself since all PDH rates can work within a relatively large rate of frequency.
One of the characteristics of these multiplexers is that the full aggregate signal needs to be demultiplexed to extract a single 2.048 or 1.544 Mbit/s tributary signal. As an example, extraction of a single 2.048 Mbit/s from a 139.264 Mbit/s requires the following operations: demultiplexing the 139.264 Mbit/s into its four 34.368 Mbit/s, then demultiplexing one of the 34.368 Mbit/s into four 8.448 Mbit/s and then extracting the 2.048 Mbit/s from one of the 8.448 Mbit/s.
Digital switches were specified to switch the 64 kbit/s time slots transported by the PDH signals; these time slots carry either voice channels or data channels. The bandwidth of these data channels was either 64 kbit/s for ETSI or 56 kbit/s for ANSI, depending on the type of signaling channel. These switches are the key elements of the PSTN.
These switches multiplex and demultiplex 30 time slots of the 2.048 Mbit/s, 24 time slots of the 1.544 Mbit/s and either decode the voice or transmit the data to the 64 kbit/s data interfaces for transmission to the access network.
On its input ports, a switch stores the incoming time slots of a 2.048 Mbit/s, or 1.544 Mbit/s, in a buffer at the rate of the input signal and on the output ports, it generates 2.048 Mbit/s, or 1.544 Mbit/s, at a rate given by its internal system clock. Buffers, or other mechanisms, are designed to allow some short-term variation in the incoming rates to be absorbed.
However, in cases where a fixed frequency offset may be present, the distance, in bits, between the read and write pointers will continually increase, or decrease. At some point the read and write pointers will collide and data will be corrupted. This situation is often referred to as buffer spill, or slip, and will occur with a periodicity based on the frequency offset and the size of the buffer used.
Of course, a very large buffer would result in rare data loss, but buffer size will ultimately impact latency and cost. Initial voice switches had buffers that allowed processing of a full PCM frame. For 1.544 Mbit/s signals buffers sizes were therefore based on 193 bits as this was a convenient size to accommodate 1 bit and 24 bytes, representing 24 voice channels of 64 kHz. By choosing the buffer size to be a multiple of 24 voice channels, a slip resulted in the loss of only one PCM voice sample, which in many cases only minimally impacted the service quality, provided that slips were relatively infrequent. For 2.048 Mbit/s signals, buffer size was based on 256 bits, representing one frame. If the PCM sample represented voice traffic, the slip was largely tolerable. However, with the growing amount of data, carried either as voice band data (or even fax) or later digital data, the impact of a slip could vary. To control performance, international standards, for example ITU-T Recommendation G.822 [G.822], defined acceptable slip rates.
The slip rate could be controlled if all switches were timed from a common source. This required a dedicated network solely for the distribution of a common clock frequency; PDH signals were commonly used to synchronize the switches. Network elements within a network would receive timing from this network. To accommodate the real possibility of failures (e.g. the so-called “Back-Hoe Fade”), intermediate clocks were developed that would minimize the drift rate during these failure conditions. Holdover performance, for example, was directly defined by the slip rate objectives necessary to enable acceptable performance of voice switches. This aspect of the synchronization networks can therefore be attributed to the introduction of digital switching.
The introduction of PSTN has created a need for the transport of frequency. Digital switches have to be synchronized in order to switch 64 kbit/s time slots transported by 2.048 and 1.544 Mbit/s signals without generating slips.
The desynchronization of a single switch in a chain of switches causes two periodic generations of slips of opposite nature, a byte canceled or doubled, in two consecutive switches, as shown in Figure 1.1, without any compensation between the two opposite slips.
The digital switches, switching 2.048 or 1.544 Mbit/s, have to be locked on a reference frequency to prevent the generation of slips. This is achieved by transporting the reference frequency in the 2.048/1.544 Mbit/s PDH signals connecting the digital switches. For this reason, it is essential that all switches of the PSTN are locked to a common reference frequency; this resulted in the development of a first synchronization network.
At this time the synchronization network was composed of a Primary Reference Clock, the PRC defined in ITU-T Recommendation G.811 [ITU-T G.811] with its well known 1.10-11 long term accuracy (with regard to the Coordinated Universal Time (UTC) as defined in G.811) allowing to limit the periodicity of slips to less than one per 70 days. The maximum drift between two switches locked to two different PRCs is twice 10–11; this causes a phase error of twice 824 ns per day or 115 µs per 70 days. A slip is generated when the drift reaches 125 µs, the period of a 2.048 or 1.544 Mbit/s frame.
Figure 1.1. Generation of slips
The delivery of this frequency reference to equipment generating a 2.048 Mbit/s signal is done via a 2.048 MHz, or even a 10 MHz, signal. It is also very frequent that the frequency reference is provided via a 2.048 or 1.544 Mbit/s signal.
The 2.048 Mbit/s or 1.544 Mbit/s signal can be multiplexed within a higher PDH rate and transported to another PSTN switch where the 2.048 Mbit/s is extracted for the PDH signal and used as a synchronization input by the clock of the switch.
Note that some switches, handling only higher PDH rates, do not need to be synchronized, due to the specific justification process used. This enables these switches to provide timing transparency to the 2.048 Mbit/s signals.
There may be many switches in a PSTN, causing noise accumulation along the chain of 2.048/1.544 Mbit/s; for this reason, a new type of high-quality clock called SSU has been defined in ITU-T Recommendation G.812 [G.812] with a very narrow bandwidth in the mHz range in order to filter the noise outside this bandwidth. This SSU ensures also other important features such as holdover to maintain the quality of the clock in case the incoming 2.048 or 1.544 Mbit/s disappears due to any failure in the network; it also distributes the reference frequency to other equipment within the node where it is installed. These clocks might be embedded in a switch directly or might be in stand-alone equipment.
To ensure robustness of the synchronization network, a switch must be able to receive the reference frequency from at least two independent inputs. This is an important task for the network operator to verify that in any failure condition, a timing loop will not be generated. As defined in section 3.3.1, a timing loop is a situation where equipment is synchronized on a timing signal it has generated. Careful network planning is required.
The transport of 2.048 or 1.544 Mbit/s signals between digital switches is mainly done by multiplexing those signals within higher PDH rates. The PDH justification process allows this multiplexing into a higher rate PDH signal without the need for the different rates to be synchronized to a common reference frequency, but it is simply needed that all these rates are within a certain frequency range specified in the ITU-T Recommendation G.703 [G.703]. This allows the transport network to remain asynchronous from the digital switches requiring reference frequency, while being the medium used to propagate the synchronization between these digital switches.
1.2.2. Evolution toward SDH and synchronous optical network
The increased demand in bandwidth and the introduction of the optical fiber and its connecting technology pushed the standard bodies, ANSI, ETSI and ITU-T, to develop a new transport hierarchy allowing an increase to the possible rates and hence the transported bandwidth. ETSI defined SDH, and ANSI defined Synchronous Optical Network (SONET).
Prior to the introduction of SDH, multiplexing systems were based on PDH multiplexing technology. Bit stuffing justification mechanisms were used to rate-adapt a lower bit rate signal for multiplexing into a higher bit rate “carrier”.
The new SDH/SONET hierarchy was defined so that it made it possible to extract and insert individual tributary signals without being forced to terminate all the tributaries, as is the case in PDH. This feature allowed the development of a new type of equipment, the Add-Drop Multiplexer (ADM), which will be the basic element for a new telecoms architecture with the deployment of rings, rather than linear lines connecting switches centers in the previous PDH and PSTN networks.
SDH was defined and standardized to alleviate some of the issues with PDH-based networking. PDH multiplex systems were generally limited to a specific type of client interface and the addition of new clients or rates usually required considerable network and equipment redesign. With the subsequent development of data networks, initially ATM, the need for flexibility was also seen in terms of the types of signals and associated bandwidth that could be carried on a uniform transport system.
SDH, while seemingly offering the same functionality as PDH systems, differed in that it offered two levels of rate justification. This was key to the flexibility required to carry various types of signals in a ubiquitous and flexible network infrastructure. Client signals were mapped into one or more “containers”. For traditional PDH client signals, this could involve a bi...
