Self-organizing networks (SON) address the
challenges posed by increasingly complex mobile networks by automating the
provisioning, configuration and reconfiguration of the radio access network
(RAN) for optimal capacity and performance. The benefits are clear. Operators
who have installed it have seen more than a 15 per cent improvement in capacity
utilization and greater than 20 per cent improvement in the dropped call rate. However,
like any powerful technology, it needs
to be put in the right hands.
To
explain the implications, we need to start at the beginning. There are two main types of SON, decentralized
and centralized. Decentralized SON (D-SON) only applies to a single node and does not allow for any coordination
between different infrastructure vendors' equipment. With
D-SON, because the SON process happens at a distributed level, its effects are
very localized. Nonetheless, that can
set up conflicts with other instances of SON elsewhere on the network. From experience, most operators that have
installed some level of D-SON on their network soon realize that they need some
level of centralized organization layer.
Because of those
limitations and scope for conflict, there is even some concern that D-SON may
already be a redundant technology.
Centralized SON (C-SON),
on the other hand, allows the whole network to be self-optimized because it is a conglomeration of multiple, smaller use
cases that focus on solving a single problem across the whole network. But even with C-SON, the vendor absolutely
needs experience of live, in-network installations of SON or risk setting up
conflicts in the operator network, which could potentially lead to making the
network worse, not better.
There are two key
functions required to prevent conflict in C-SON. These are the trigger
function, which determines which use case to initiate based on specific
symptoms, and the coordination function, which manages any conflicts on the outbound
command path to the network. The coordination function plays a pivotal role in
deciding which command should take precedence over another, and in which order
they should be transmitted.
Without the
coordination function in place, a worse-case scenario of “Runaway SON” could
occur. It would come about if a command
gets into a never-ending loop between two cells, setting in motion a chain of
events which can ultimately lead to network outages. For example, one bad
command may create a condition on one cell that the SON system senses, and
corrects it on that and subsequent cells. Then, the first SON system catches
the “bad” condition on the “corrected” cells, amplifying a single mistake and
catalyzing a chain reaction of errors.
While this is currently only a theoretical scenario, for SON vendors
without in-network SON experience and years of network management experience,
it is a potential reality.
Mobile operators are
sometimes wary of autonomous systems because of the risk of such outages
occurring without human involvement. But humans do not have to be completely
left out of the picture, particularly at the early stages of an
installation. Operators can start by
incorporating some human involvement into the command loop. This “open-loop”
method allows for greater control of the SON system since live personnel can
decide if a particular change to a cell should be pushed to the network or not,
minimizing the risk of a command being incorrectly prioritized and causing
network outages. Once the foundation of trust is built, human intervention can
slowly be removed from the equation to enable a completely autonomous SON
system, once real people have seen how effective the system is.
Operators that have
installed an “open-loop” system have made the switch to fully autonomous in as
little as four weeks in our experience.
This shows just how quickly operators can be convinced of the real-world
benefits of SON when they see it in action.
If operators keep a skeptical eye on D-SON and do not rush to engage
partners without live SON experience, there is no need for conflicts to ever
become an issue. Then finally, the
positive impact of SON, and C-SON in particular, of improved customer
experience and reduction in CAPEX and OPEX can be felt universally.
In the past two years, Amdocs entered a new space of managing networks' traffic by acquiring two companies - Celcite (2014, $129M, here) and Actix (2013, $120M, here). See also "Amdocs Integrates RAN Management and Optimization Solutions" - here.
Answer by Eli Gelman [pictured], President and CEO, Amdocs: "So network software in general, including network optimization, is moving along well. Its lumpy, because the deals there are not so smooth I would say in terms of the deals, the deal flow, not the execution.
Posts
