• Umut Keten - Lead Architect Networks, Turk Telekom

A major part of the 5G evolution is meeting the phase and frequency demands. Networks based on time division duplex (TDD) and C-RAN techniques.

Operators are trying to design networks capable of delivering these demands. 5G base stations require extreme accuracy, from a few hundred (coordinated) nanoseconds to a maximum of 1.5 microseconds.

With these requirements for synchronisation in mind, the use of PTP for phase and time distribution was developed by IEEE.
The PTP telecom profile [ITU-T G.8275] defines the parameters from [IEEE 1588] to be used in order to guarantee protocol interoperability between implementations and specifies the optional features, default values of configurable attributes and mechanisms that must be supported. However, it does not guarantee that the performance requirements of a given application will be met. Due to this ‘unguaranteed’ behaviour and due to cost and limitations on existing network elements operators need to locate multiple grandmaster clock technology close to small cells and end applications.

Also due to these unguaranteed behaviours, operator networks are relying on the Global Navigation Satellite System (GNSS) to deliver Coordinated Universal Time (UTC), the main issue and challenges here are outages. The band signals are very weak it is easy to jam and not to forget bad weather conditions can also a be a common reason for outage. But most importantly Geo-Politically some countries will not want to be relying on GNSS technology that in some cases is military controlled.

The purpose of this paper is to present our new method which we have tested successfully on our live WAN network. Our method is a more functional, less complicated (in terms of implementation, operation), more robust (tolerance to jitter/pdv, asymetry, link loads, hop count and convergence) and highly accurate Time Transfer architecture over IP/MPLS networks without the use of current ITU-T adaptations [G.8275] Full Timing or Partial Timing Support. All of the robustness is accomplished by the use of DTM (Dynamically Synchronous Transfer Mode) protocol as described in ETSI ES 201 803.

The test results are something that the PTP telecom profile [ITU-T G.8275] never can achieve. Our testing was done on our live MPLS network with live traffic, as noted earlier without any Timing Support or RSVP-TE, an ordinary VPN service was assigned with AF42 QoS parameters and a 25mbps bandwidth.

14 nodes without timing support and legacy parts was traversed from Adana,TR to Istanbul,TR, up around 1200Km of fiber and asymetry was present. The PRC received in Adana was transported to a EnodeB 5G site in İstanbul and compared to the GPS reference in İstanbul. An average of 130ns offset was measured with lows around 50ns and peaks at 290ns. Fiber cuts and reroutes in the MPLS network happened naturally but also cuts and reroutes where triggered manually, despite these network changes the delivered Sync was extremely accurate. Currently more 5G sites are being deployed and the results will be presented at WSTS.