Timing is part and parcel of edge computing owing to particular network and service requirements at the edge. The biggest driver of edge computing is latency reduction. This reduction is achieved by running services closer to the user. However, latency reduction is not the only driver. More and more networks and services have come to depend on strict and accurate time requirements.
Such as:
- Financial markets require strict accurate timing as it depends on how fast and accurately users execute transactions
- Industry 4.0-based automation requires sophisticated and fast control loopback analysis and actions from the data collected from different sensors.
- 5G requires stringent time base timing requirements versus 4G, which requires less stringent frequency requirements.
Any degradation in timing will result in poor QoS, which translates into a bad customer experience.
Fortunately, IEEE 1588 is one such standard that can utilize the network to distribute highly accurate clock information from a hub site. In this case, the hub site is the edge computing site that can provide an ideal location for deploying a 1588 clock source.
What is IEEE 1588 ( PTP)?
IEEE 1588 ( also called Precision Time Protocol- PTP) provides a highly accurate and reliable clock mechanism that uses packet technology to transfer clock/synchronization information from a source site that has Grand Master (GM) to the user sites ( also called clock slaves)
The V1 version of the 1588 standard defined clock accuracy for industrial use cases which was revised through version 2 in 2008, which provided enhanced accuracy for the communication industry, also called 1588 v2-2008
Isn’t the Packet transport unreliable for carrying clock information?
Yes, It is!
The packet network is non-deterministic as it is prone to jitter and delay. It seems unsuitable to carry deterministic information such as synchronization packets.
However, PTP packets are carried at layer 2 combined with the hardware timestamp at layer 1 ( which is the physical layer), thus making it highly reliable, predictable, and accurate.
Different types of profiles for 1588 in edge computing
The profiles provide benchmarks, limits, and deployment options for different industries like telecom, industry, power, etc.
For example, ITU-T is very active in defining profiles for the telecom industry, while IEEE has defined profiles for the power industry, such as:
Telecom Profiles
Some of the notable ones are
G.8265.1 Telecom Profile for frequency distribution.
G.8275.1 Telecom Profile for time/phase distribution using full timing support.
G.8275.2 Telecom Profile for time/phase distribution using assisted partial timing support.
Industrial/Power Profile
IEEE C37.238-2011 is a Power profile, Grandmaster, Boundary Clock and Ordinary Clock for use cases related to industry and power (smart grid)
Let’s zoom into different verticals to see the use cases and timing requirements.
Telco Edge and IEEE 1588
Perhaps the most significant use case for 1588 is in the communication industry.
The requirement for packet synchronization has increased with the advent of 5G. In fact, there was always a requirement for frequency synchronization in 2G, 3G, and 4G for smooth handover between macro cells. Still, all these technologies needed frequency synchronization, which is less stringent. These technologies were based on FDD (Frequency division duplex)
5G, on the other hand, is TDD (Time division Duplex) based technology that uses the same frequency on uplink and downlink. The uplink and downlink are separated in the time domain. Therefore, 5G needs strict time and phase synchronization. Any issue in time synchronization results in severe interference causing QoS issues. This situation can worsen in a scenario where many small cells are deployed and thus requires careful planning of timing synchronization.
The absolute time accuracy requirement for 5G base station is around 1.5 microseconds.
But that’s not all
5G needs additional relative synchronization, depending on using advanced options in 5G like carrier aggregation, COMP, Massive MIMO, etc. The relative time synchronization between the base stations can go as low as 65 nanoseconds which stresses the synchronization accuracy to the limits
This raises a question.
Can GNSS (Global Navigation Satellite System) that can provide a time of the day and phase synchronization at every base station solve the issue?
Is GNSS-based sync, such as GPS, solve synchronization issues?
While GNSS-based synch can expedite the time to launch 5G service concerning ease of installation. They do come with specific challenges as not having a sufficient backup. A better option is IEEE 1588 combined with GNSS.
IEEE 1588 at edge site solves synch issues
Let’s take the case of the radio access network in a Centralized RAN (CRAN), where multiple remote radio units (RRUs or RUs) terminate at a specific DU site. In the following case, the operator has decided to install a GPS receiver at every RU site, thus providing time and phase locally to each base station.
But in this case, with a failure of GPS at any of these sites, the services at that site will be severely affected.
Secondly, as cell sites are synchronized separately because each GPS receiver works in stand-alone mode, it will be challenging to achieve related sync limits between the base stations in nanoseconds if advanced services such as carrier aggregation and COMP are used.
Fig: CRAN dependent on GPS alone
A better approach is to combine 1588 PTP with GPS and use the Edge site as the primary synch with the base station as the backup sync.
Fig: CRAN with 1588 as primary and GPS as backup
In this approach, the 1588 Grandmaster clock (GM) is used at the Edge site as a primary source, while GPS at the cell sites is used as a backup source.
This architecture has the benefits of better redundancy while at the same as one GM clock is serving multiple base stations, the relative frequency requirements are also met without any concern.
IIoT and Industrial Edge Synchronization
Industry 4.0 is revolutionizing manufacturing with modern technologies such as IoT, cloud computing, AI, industrial automation, and machine learning. These technologies allow for improved decision-making with advanced sensors, embedded software, and robotics that collect and analyze data.
However, it is essential to have alignment in the time of the data that come from different sensors. For big data to provide accurate data analysis, the time stamps must be accurate enough. Any incorrect data here has the potential for wrong judgments here. The control systems have to deal with a lot of real-time processes where IIOT devices are reporting the data to control loop systems. Traditional control systems are typically scan-based or asynchronous, which can suffer from processing jitter. However, time-based control using 1588 can provide better performance and must make faster decisions, thus mandatory for industrial automation.
Lanner’s HTCA Platform with sync functionality
Lanner’s hyper-converged all-in-one MEC HTCA platforms HTCA-6600 and HTCA-E 400 provide switch blades to support PTP/ 1588v2, complying with ITU T G.8265.1 and G.8275.1 profiles, suitable for diverse applications such as RAN (CU, DU) CDNs, Anti-DDoS NGFW, BRAS, and MEC.
HTCA-6600 has a high availability Chassis 6U size with 6 x86 CPU Blades and 6 I/O Blades.
HTCA-E400 is a compact-size platform that supports 5x 1U compute sleds or 2x 2U+1U compute sleds with 450mm Short Depth suitable for Edge Deployment.
Lanner operates in the US through its subsidiary Whitebox Solutions (whiteboxsolution.com)
