Network resilience
32
When a master fails, the
system will automatically make
www.criticalcomms.com March 2020
site is very remote, although in the Middle East we have
seen a weekly replenishment commonly.”
In terms of security, Humphries adds: “Perimeter
security can be enhanced with IR sensors and CCTV
cameras, and it is also helpful to have CCTV in the shelter
to see who has access to the cabinets. Alarms on doors
or fences and gates can alert network managers to
unauthorised ingress.”
Vicente Rubiella, product manager – systems, NEBULA
TETRA Infrastructure at Teltronic, adds: “Back-up powergeneration
types will depend on the customer preferences
and terrain characteristics. For example, solar panels can
be a good option for isolated areas where grid connection
is impractical or involves an unreasonable cost. Backup
power-generation equipment must be dimensioned
accordingly to provide power throughout the expected
time to restore the primary power source in case of failure,
usually 24 hours.”
Jochen Boesch, senior director, engineering at Damm
Cellular Systems in Denmark, states: “Typically -48Vdc
is used to power base stations. This allows for generator,
batteries and solar panels in parallel usage. Our base stations
also offer 230Vac input to use a UPS.”
Humphries says: “Critical sites should have duplicate
links, perhaps from different suppliers. Different
technologies can come into play here. If two separate fibre
or wire links cannot be provisioned then a microwave or
satellite link may be used. Remote sites may only be able
a slave node the master
to use satellites; good practice would be to connect
through two different satellites. 4G gives us another link as
a back-up.”
Switch and core
There are different approaches to centralised or distributed
switching. Motorola Solutions favours centralised switches.
It cites the reduced traffic between the Zone Controller
which tracks all base stations, compared with a distributed
system where all base stations need to negotiate and pass call
information to each other.
This zone controller also knows where every radio and
group are located and only sets up calls on base stations
where users are present. In terms of resilience, services are
maintained if there are local failures. Humphries adds:
“Although centralised, our systems provide duplicate
processors and there can also be mirrored processors in
different locations to provide geographical redundancy.
Regional centres such as the counties in the UK can
take over the control in another county in the event of a
major failure.”
Boesch from Damm makes the case for distributed
switching. “Distributed networks can keep a full feature
set if a subset of nodes become isolated, and are easily
scalable with new sites added without increasing switch
capacity. As well as quicker set-up times for local calls, the
backbone traffic can match or better a centralised switch if
set up intelligently. A master/slave concept for back-up in a
decentralised system is easy to handle. When a master fails,
the system will automatically make a slave node the master –
the same also goes for applications connected to gateways.”
Teltronic advocates centralised switching. Asked whether
this is less resilient than distributed, Rubiella states:
“Definitely not. Centralised architectures provide all kinds
of facilities to feature the highest level of resiliency, when
redundancy is vital. Having a centralised switch and a hot
standby back-up switch, where no user action is needed to
perform an immediate switchover on detection of failure
condition, achieves the most resilient network deployments
in a cost-effective way.”
Regarding failover mechanisms, Rubiella concludes:
“Failover must rely on two basic and essential requirements:
rapid detection of failure condition to trigger switchover
immediately; and no user action needed, but automatic
switchover between main and back-up unit.”
This debate will continue. Potential users would be wise
to carefully compare offerings from several suppliers to make
the best decision for themselves and their particular needs.
LTE joins the club
The introduction of mission-critical specifications into
3GPP 4G and now 5G means cellular network operators
also have to consider higher levels of resilience. With the
3GPP Releases 12-14, public safety features were added
to the LTE standard including Proximity Services. This
Adobe Stock/fotosr52
Opinions differ
on the merits of
centralised and
decentralised
network
architectures, but
measures can be
implemented to
make both highly
resilient
/www.criticalcomms.com