Instrumentation & Measurement Magazine 25-6 - 13

explained in the previous Section. As shown in Fig. 1a and Fig.
1b, an SAS consists of multiple bays (called also process buses)
whose goal is to control the field-level devices (e.g., circuit
breakers, relays) of a portion of a line by using the data collected
from various sensors and IEDs. A local network allows
all of the IEDs in a bay to communicate. The station bus connects
the various bays of an SAS to the SCADA workstations.
A Global Positioning System (GPS) receiver is typically
used to satisfy the most stringent synchronization requirements.
Indeed, a GPS receiver can provide not only accurate
estimates of the position of an object, but also an accurate time
reference (i.e., in the order of ±100 ns) as a result of the frequency
stability of the Cesium clocks in the GPS satellites. The
GPS receivers usually include a dedicated 1 Pulse Per Second
(1-PPS) signal that can be used as a local time reference. The
main disadvantage of GPS-based solutions is that they are not
scalable, as they need an antenna for each receiver and, therefore,
for each network node. However, the time obtained from
a GPS receiver can be distributed, even indoors and over long
distances, using the Inter-Range Instrumentation Group Time
Code B (IRIG-B) over a dedicated copper or optical fiber cable.
Fig. 1a shows how the time reference signal obtained from
a GPS receiver is typically distributed to the bays of a primary
substation. It is worth noting that the IRIG-B clock distribution
network (dotted lines) relies on a physical infrastructure that is
different from the main communication network (solid lines).
This solution is widely used in the power industry. If the delays
due to the signal propagation through the IRIG-B cables is
compensated, a time synchronization accuracy better than 1 μs
can be easily achieved. Quite importantly, IRIG-B-based solutions
do not support any redundancy mechanism different
from physical network duplication.
Network-based Time Synchronization
Although the synchronization solutions based on the distribution
of a signal through a dedicated wiring infrastructure
provide good accuracy, they lack flexibility and they suffer
from high installation costs. However, the time information
can be alternatively distributed via a communication network,
as shown in Fig. 1b. In this case, the information on
time, provided by a GPS receiver, is encapsulated in messages
and broadcasted from one or more reference nodes (denoted
as a Grandmaster in Fig. 1b) to the rest of the networked devices.
The network delay used to transfer the time information
is measured and compensated by a synchronization protocol.
Thus, any error in estimating this contribution may have
a significant impact on the overall accuracy. Unfortunately,
the network delay is a non-zero-mean random variable that
depends on several contributions (e.g., the network load or
possible cables and transceivers asymmetries).
The time reference node and each device to be synchronized
exchange a sequence of timestamped messages to
estimate the network delay and the time offset between them.
One of the main sources of uncertainty in network-based
synchronization is indeed timestamping. Timestamping accuracy
depends indeed on the layer of the ISO/OSI stack of
September 2022
protocols at which the ingress times of incoming messages and
the egress times of outgoing messages are actually measured.
Low-level timestamping is usually more accurate.
The Network Time Protocol (NTP) was designed to distribute
time over networks with variable propagation delays by
using TCP/IP messages. The nodes of an NTP network, called
clocks, are arranged into a hierarchical structure. A single NTP
clock can recover time information from one or more redundant
time servers: the " Clock Selection Algorithm " decides
which time references should be selected based on their quality.
The NTP synchronization accuracy is in the order of a few
ms, i.e., sufficient for the power systems applications of Classes
C, D, E, F (as shown in Table 1).
The Precision Time Protocol (PTP), standardized as IEC/
IEEE 61588-2021 or simply IEEE 1588 [2], is a time synchronization
protocol similar to NTP, but it is optimized for Local
Area Networks (LANs). The PTP protocol usually self-arranges
the clocks of the devices connected to a LAN into a
tree structure with a master-slave hierarchy: the clock with
the highest stability is automatically selected by the Best Master
Clock (BMC) algorithm as the local time reference, also
called Grandmaster clock. The Grandmaster clock sends multicast
synchronization messages to the rest of the nodes, which
progressively estimate the message propagation delays and
their time offset with respect to the nodes at a higher level of
the hierarchy, so that all nodes are ultimately synchronized
to the same Grandmaster clock. Local servo-clocks are used
to discipline the tick rate of slave nodes. PTP software-only
implementations timestamp all messages at the operating
system level (i.e., within the device drivers). This solution
ensures full compatibility with existing and non-PTP network
devices but at the price of a lower accuracy (generally
in the range 1-10 μs). Instead, the PTP solutions, in which all
devices involved in a time transfer chain perform message
timestamping at the physical or Media Access Control (MAC)
layer, can ensure sub-synchronization accuracy. In this case,
fully PTP-compliant switches are required. They can be classified
as Boundary Clocks (BCs) or Transparent Clocks (TCs).
The former ones are used in hierarchical networks and include
servo-clocks as well, whereas the TCs are primarily
conceived for networks consisting of many cascaded devices
(as it frequently happens in power systems applications) because
they just measure and compensate for the propagation
delays of all the PTP messages crossing them.
The PTP profiles describe specific settings of the PTP protocol
conceived to optimize synchronization performance in
specific application domains. Among them, the Power Utility
Profile (PUP), standardized as IEC/IEEE 61850-9-3 [3],
is focused on the needs of the power industry. The key PTP
settings specified by the PUP are: hardware-only PTP implementation;
use of Layer-2 synchronization messages over
Ethernet; support of priority mechanisms for communication;
fault tolerance through redundant synchronization messages
over duplicated network paths; support of simultaneously active
redundant master clocks; and use of TCs only as network
switches. As a result, a PTP implementation based on the PUP
IEEE Instrumentation & Measurement Magazine
13
Instrumentation & Measurement Magazine 25-6

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-6
