Self organising networks will be crucial to the efficiency and cost-effectivess of mobile broadband networks. But will they work?
Self Organising Networks (SON) is a new technique being introduced alongside LTE/SAE as part of the next generation mobile broadband network technology, and is endorsed by the NGMN alliance as a key requirement for future networks. The objective is to automate the configuration and optimisation of base-station parameters to keep best performance and efficiency. Previously a drive test team would go out into the live network and take a ‘snap-shot' of the performance, then bring this back to the lab and analyse to improve the settings. Of course more snap-shots give more data and hence better optimisation, but this data acquisition process is expensive, difficult, and not repeatable. In addition, this is a reactive method to cure problems after they have occurred, and this does not help improve customer experience.
SON will enable network operators to automate these processes using measurements and data generated in the base-station during normal operation. By reducing the need for specific drive test data, this technique should reduce operating costs for an operator. By using real time data generated in the network, and reacting in real time at the network element level, this should enhance customer experience by responding more dynamically to changes and problems in the network much earlier so that users are less affected.
Basic principles
SON is the top level description of the concept for more automated (or fully automated) control and management of networks, where the network operator has only to focus on policy control (admission control, subscribed services, billing etc) and high level configuration/planning of the network. All low level implementation of network design and settings is made automatically by the network elements. The self organising philosophy can then be broken down into 3 generic areas relating to the actual deployment of the network, these are configuration (planning and preparation before the cell goes live), optimisation (getting the best performance from the live cell), and healing (detection and repair of fault conditions and equipment failures). Each of these is further explained below.
Self Configuration
This is the first stage of network deployment, and covers the process of going from a ‘need' (e.g. need to improve coverage, improve capacity, fill a hole in coverage etc.) to having a cell site ‘live' on the network and providing service. The stages involved here are roughly:
– Planning for location, capacity and coverage.
– Setting eNb parameters (radio, transport, routing and neighbours).
– Installation, commissioning and testing.
The Self Configuring network should allow the operator to focus on selecting location, capacity and coverage needs, and then SON should automatically set the eNodeB parameters to enable the site to operate correctly when powered on. This will in turn minimise the installation and commissioning process, and enable a simple "final test" at the site to confirm that the new site is up and running.
Self Optimising Once a site is live and running, there are often optimisation tasks to be made that are more of a ‘routine maintenance' activity. As the geography of the area changes (e.g. buildings constructed or demolished), the radio spectrum changes (e.g. new cells added by the operator or by other operators, or other RF transmitters in the same area or at the same tower), then the neighbour cell lists, interference levels and hand-over parameters must be adjusted to ensure smooth coverage and handovers. Currently, the impact of such issues can be detected using an OSS monitoring solution, but the solution requires a team to go out in the field and make measurements to characterise the new environment and then go back to the office and determine optimum new settings. SON will automate this process by using the UE's in the network to make the required measurements in the field and report them automatically back to the network. From these reports, new settings can be automatically determined. This will remove the need for drive test teams to make such measurements. This concept can also be extended to managing Quality of Service and load balancing by using quality reports to optimise the scheduling algorithms in the eNodeB.
Self Healing (fault management and correction)
When a site is fully operational and active then it is generating revenue/satisfying customers. If there are any problems with the site and it fails to provide a service/coverage then revenue and/or customers are lost, and so the site must be brought back up to full capacity as soon as possible. The third element of SON is to automatically detect when a cell has a fault (e.g. by monitoring both the built in self test, and also the neighbour cell reports made by UE's that are/should be detecting the cell). If the SON reports indicate a cell has a failure then there are 2 necessary actions; to indicate the nature of the fault so the appropriately equipped repair team can be sent to the site, and then to re-route users to another cell if possible and to re-configure neighbour cells to provide coverage in this area whilst the repair is underway. After the repair, SON should also take care of the site re-start in a similar process to the site commissioning and testing.
Technical issues and impact on network planning. To deploy SON in a multi-vendor RAN environment requires standardisation of parameters for reporting and decision making. The eNodeB will need to take the measurement reports from UE's and also from other eNodeB's and report them back into the O&M system, to enable optimisation and parameter setting. Where there is multiple vendor equipment involved then this must be in a standardised format so that the SON solution is not dependant on a particular vendor's implementation.
The equipment vendors who are implementing SON will need to develop new algorithms to set eNodeB parameters such as power levels, interference management (e.g. selection of sub-carriers), and hand-over thresholds. These algorithms will need to take into account the required input data (i.e. what is available from the network) and the required outcomes (including co-operation with neighbour cells).
Further more, as SON is also implemented into the core network (Evolved Packet Sub-system, EPS), there needs to be standards on the type and format of data sent into the core. Inside the core network, new algorithms will be required to measure and optimise the volume/type of traffic flowing taking into account the Quality of Service and service type (e.g. voice, video, streaming, browsing). This is required to enable the operator to optimise the type and capacity of the core network, and adjust parameters such as IP routing (e.g. in an MPLS network), traffic grooming, etc.
Effects on network installation, commissioning and optimisation strategies
Equipment vendors have the opportunity now to develop algorithms that link eNodeB configuration to customer experience, allowing fast adaptation to customer needs. The challenge is to link RF planning and customer ‘quality of experience' closer together at a low level technical implementation. The benefit is that the network can adapt to meet the user needs in the cell without additional cost of optimisation teams constantly being in the field. The network planners' simulation environment will now need to take into account the SON operation of eNodeB when the making simulations of capacity/coverage for the network. As the operator may not directly control/configure the eNodeB, the simulation environment will need to predict the behaviour of the network vendors SON function in the network.
Operator/Installer's site test must verify that all parameters are correctly set and working in line with the initial simulation and modelling. This will ensure that the expected coverage and performance is provided by the eNodeB. The SON function will then ‘self optimise' the node to ensure that this performance is maintained during different operating conditions (e.g. traffic load, interference). This should reduce the amount of drive test required for configuration and optimisation (in theory reduced to zero), and drive test is only needed for fault finding (where SON is not able to self heal the problem).
Conclusion
SON can simplify operator's processes to install new cell sites, reducing the cost/time/complexity to install new sites. SON gives an obvious benefit if deploying femto cells, as the operator is not strictly in control of the cell site and needs to rely on automated processes to correctly configure the cell into the network. In addition, the running costs of the site are reduced, as drive test optimisation is reduced and site visits for fault investigation and repair can be reduced. All of this leads to OPEX savings for the network by using automated technology to replace manual operations.
For the customers on the network, SON will lead to better customer satisfaction as coverage and Quality of Service are driven and optimised by actual customer usage, and there should be reduced downtime or faulty cells. The OSS monitoring systems and SON should work together to automatically detect usage trends and failures and automatically take action in real time to correct errors.
About the author:
Jonathan Borrill, Director of Marketing, Anritsu EMEA Ltd.