News & Analysis
Synching Up Networking Equipment Designs: Part 2
James Porter, Datum-Austin
8/14/2002 4:34 AM EDT
With carriers looking to map a slew of services, including IP and ATM, over existing Sonet links, synchronization becomes even more important. With that in mind, design engineers must engineer their networking system architectures properly in order to promote proper network synchronization.
This is second installment in our series on designing/implementing synchronization techniques for networking equipment designs. In Part 1, we looked at the key timing components and standards required to make synchronization come to life. Now, in Part 2, we'll look at the key components and design consideration that designers must use to build a synchronization system for networking equipment designs.
Basic Systems Components
As Figure 1 shows, a synchronization system can be broken down into five components.
An input system lies at the front of the system is an input section. This section accepts clocking signals and translates them into a common format. For example, if input signals are in DS1 or E1 formats, the input section extracts the clock signal and translates it to an 8 kHz signal. The common 8 kHz clock signal is then used by the other components of the systems.
In addition to format translation, the input section monitors performance and passes along sync status messages (SSMs) and other messages to the rest of the system. Performance measurements include maximum time interval error (MTIE) and time deviation (TDEV) calculations on the input signals.
The timing section is the heart of a synchronization system. The timing block takes the common clocking signal from the input section and generates a stable and accurate clock signal using a phase-lock loop or frequency-lock loop. The timing section also filters out unwanted jitter and wander from input signal and is responsible for selecting the best clock signal available at its inputs. The timing section's ability to switch to a secondary input clock provides redundancy, which is necessary to meet the high-reliability requirements synchronization system need.
The timing section uses the oscillator reference output within its phase- or frequency-lock loop section. The reference section provides a stable oscillator signal and stability during holdover mode. In the event that all primary-clock inputs fail, the timing block will use the holdover reference signal until repairs re-establish a primary clock. The holdover reference usually comprises a Stratum 3 or 3E crystal oscillator or may even use a Stratum 2E rubidium oscillator.
The output section translates the synchronization system's common clocking signal to the desired output format. Some common output formats include DS1 or E1, or telecom- or OC frequency clocks. The output section also translates messages to the appropriate format.
The fifth component, the communications section, allows the synchronization system to interface to its host system. Through the communications section the user can configure the system and access system status, alarms, and other messages. The communication block also allows you to monitor the system's performance. The user interface might be a serial connection or a parallel, register-based interface. Finally, to tie all the sections together and enable the system to operate properly the individual system blocks must be able to communicate with one another through a common bus and communications protocol.
Redundancy Needed
When designing a synchronization system, redundancy is one of the key considerations. Due the high-reliability requirements of today's network operators, some method of ensuring continued operation in the event of the loss of an external reference signal must be included in any synchronization design. The two most common methods involve a backup reference source.
In the first method the input section has inputs for more than one reference signal. In normal operation the input section uses its primary input as a reference source. If the primary source fails or its quality falls below that of a secondary clock signal, the input section switches to the secondary clock either automatically or under manual control.
The second method of supplying redundancy is based on the master-slave nature of networking clocks. If all external reference sources fail, the system's timing section switches over to the holdover reference's internal oscillator signal, or slave clock. The accuracy of the slave clock is always less than or equal to the master clock. The length of time a synchronization system can remain in holdover mode depends on the stratum level of the slave clock.
Whether you are designing the synchronization system's basic components from scratch or buying off-the-shelf components, designers' ultimate goal is to build a clock system that meets the appropriate standards specifications. Designers should be concerned with specifications from ANSI, ITU-T, ETSI, and Telcordia. Specifications from these bodies describe the performance of the overall system and not the requirements of individual components.
Critical Design Considerations
Designing network element synchronization is no easy task. Those who do it for a living have spent years acquiring their expertise. There are many design issues to consider. These issues are often complex and hardly intuitive.
While overall system specifications are spelled out in the documents just mentioned, It's seldom obvious what specifications you must meet when designing the basic synchronization components. To make matters more difficult, synchronization design isn't something you can pick up from a textbook or class. It's something you learn through years of experience, hopefully with the help of someone who has gone down this road ahead of you. But that's not likely because there are very few synchronization experts around.
To ensure that your design meets the low phase-distortion requirements needed by network element, design engineers need to carefully select components with low noise characteristics. Keep in mind that each component can add noise to your system in a cumulative fashion. Unfortunately, selecting and purchasing appropriate synchronization components is only about 20% of the design problem. The other 80% involves integrating the components to meet the system-level specifications. This effort can amount to hundreds of man hours of engineering effort.
Before embarking on a project to design your own networking synchronization system, ask yourself if you have the expertise in house to even begin such an effort. Don't underestimate the value of synchronization design experience. If you don't already have the engineering resources, you will have to acquire it, and synchronization engineers are a rare breed. If you decide that it is worthwhile to develop your own synchronization talent, ask yourself how much time you will be adding to your time-to-market. Gaining synchronization expertise is expensive and doesn't guarantee success.
Design Approaches
Strategies for building synchronization equipment can be broken down into three categories: the home-brew approach, the partial or hybrid approach, and the off-the-shelf approach (Table 1).
| Design Strategies | Time To Market | Cost | Risk |
|---|---|---|---|
| Home brew | High | High | High |
| Hybrid | Medium | Medium | Medium |
| Off-the-shelf | Low | Low | Low |
In the home-brew approach, engineers design everything from scratch. This strategy might be appropriate for companies with in-house synchronization expertise and who want as much control as possible over their design. The tradeoffs are increased time-to-market, higher cost, and greater risk. Companies lacking synchronization-design engineers or who want to focus on their core competencies will probably avoid the home-brew approach.
The second strategy is to source your synchronization components from different vendors and integrate them into your custom design. This approach also requires in-house synchronization experience and can add hundreds of man hours to your schedule.
Another drawback to the hybrid approach is you end up with a solution that addresses a specific set of requirements rather than a general-purpose design that can be reconfigured for use in other products. And the hybrid strategy doesn't guarantee that you end up with a system that meets the necessary system-level synchronization specifications. Instead, the engineer pulls together components from different vendors and hopes the resulting design meets all the specifications mentioned earlier.
Within the last few years there has emerged a third option for designing synchronization equipment—the off-the-shelf approach. A host of companies now offer ready-to-use synchronization components that are guaranteed to meet the necessary system specifications when configured together.
There are several advantages to this methodology. First, you can deal with a single supplier for all the components you need. Using off-the-shelf components eliminates the need to understand the intricacies of synchronization design. Relieving yourself of this design burden saves hundreds of man hours. Instead, you can concentrate on adding value to your networking equipment such as unique features, port density, lower cost, or higher reliability.
Much of the design risk is eliminated with this approach as well. Since all the components are designed to create a very low-noise and ready-to-use system, you can concentrate on adding features instead of debugging a poor design. Time-to-market is also reduced, which is key to any product's success.
Wrap Up
That concludes our two-part tutorial on synchronization techniques for networking equipment designs. To view part 1 of this art, click here.
About the Author
James Porter is a senior systems design engineer at Datum-Austin. He has more than 20 years of experience in the areas of time, frequency, and synchronization. James can be reached at jporter@datum.com.
|
|
|
|
|
