LTE self-organizing networks activities MERA follows communication technology evolution trends and invests in expertise in a new generation of wireless communication standards, including LTE and LTE Advanced. In 2010 MERA and Fraunhofer Institute for Open Communication Systems (FOKUS), an R&D facility located in Berlin specializing in multi-domain networks and interoperable, user-centered solutions for open communication systems, have launched a joint project in the Evolved Packet Core (EPC) domain. Since then MERA has been developing the OpenEPC (Open Evolved Packet Core) platform, e.g. Mobility Management Entity (MME) and relevant MME components and interfaces. The company also contributes to prototyping the platform components and connecting LTE (Long Term Evolution) components to OpenEPC to create more flexible and fast communication networks. One of the most important directions of MERA's LTE expertise development is Self-Organizing Networks (SON). To enhance its expertise in LTE SON, MERA is developing LTE system-level simulator, which is a Matlab-based simulator allowing to evaluate system performance in terms of given metrics subject to variation of different aspects such as LTE-specific PHY level parameters, propagation channel models, users' movement models etc. The following picture shows an example of enodebs layout and corresponding SNIR distribution: The one of the most important use cases of SON functionality is the handover (HO) parameter optimization. The HO is a procedure ensuring that users can move freely through the network, and its main stages (in LTE system) are shown in the following picture:
It is evident that HO performance is affected by different factors: Combination of HO triggers (control parameters): o Handover hysteresis, o Time to Trigger (TTT). Handover measurement accuracy: o Limited number of reference symbols available, o Different approaches to L1/L3 filtering can be applied. Errors in transmission of control information on L1 control channels: o UL: measurement report or handover confirm, o DL: handover command. Processing Delays: o UE RRC Processing Delay, o X2 Latency. The choice of appropriate values of HO control parameters influence the HO performance greatly. At the same time, adaptation of the HO triggers on a basis of
users' speed, propagation conditions, cell sizes etc. is needed. The following picture depicts the interrelation between control parameters and HO process: The main objective of SON functionality concerning the HO is to reduce OPEX by avoiding manual tuning of the HO parameters and replacing it with automated optimization. In order to optimize the HO performance, the corresponding metric (an objective function) should be put under consideration. The objective function (OF) may include one or more key indicators. These indicators, in accordance with the SOCRATES project, can be introduced as: Handover failure ratio the ratio of failed handovers to the total number of handover attempts, Ping-pong handover ratio the ratio of handed over calls that are handed back in less than critical time to the total number of handovers, Call drop ratio the ratio of existing calls that are dropped before they are finished to the number of calls that were accepted by the network. If the values of handover hysteresis and time to trigger are the same for all enodebs in the network, the objective function takes a form of 2-D function and can be shown graphically. The corresponding result obtained with MERA s simulator is shown in the following picture:
The lower the value of the objective function the better the HO performance is. As it is shown above, a ditch of low values is noticeable lying in a circular shape (magenta line) around the point with a hysteresis of 0 db and a TTT of 0 s. Hence, the optimal set of parameters does exist and the handover optimization algorithm has to drive parameters towards the area of good handover performance. Therefore, the crucial issue of SON functionality involving the HO performance is the optimization algorithm. It should find the optimum set of control parameters under the disturbances due to the measurement errors and the finite observation time subject to keeping compution complexity at the reasonable level. MERA is carrying out intensive research in this area and will be ready to propose effective solution in the nearest future. Another important use case of SON functionality is the Cell Outage Management. It combines the Cell Outage Detection and Compensation mechanisms to provide automatic mitigation of an enb failures in the cases where the enb equipment is unable to recognize being out of service and has therefore failed to notify OAM of the outage. The main goals of Cell Outage Management are: to ensure that the operator will know about the fault before the end user will be out of service subject to the minimum level of human control and intervention (Cell Outage Detection);
to alleviate the problem caused by the loss of a Cell from service by means of automatic reconfiguration of the neighboring cells (Cell Outage Compensation). The Cell Outage Compensation task can be solved by the regulating of a cell s coverage footprint. This can be performed in different ways: by tuning of the handover parameters; by adjustment of the antenna parameters (azimuths, titles, beamshapes) or/and the transmit power. Mera s LTE system-level simulator allows to model the Cell Outage Compensation process and to see it in the progress. The user of simulator is to set the enb to turn off and the corresponding network state is fixed as the initial point of simulating. The simulator assumes that the network state can be characterized by the single indicator the value of an M-ary objective function (OF). The OF is the user-defined and, as a rule, is composed of several partial metrics. The typical metrics are: the distribution of Signal-to-Noise and Interference Ratios and the corresponding derivatives; the Beam Loading (ratio of the number of users which are served by the least loaded enb, to the number of users, which are served by the most loaded enb). For any network state the simulator tries to find the best (extreme) value of the OF by adjustment of its variables (azimuths, titles, beamshapes, transmit power) subject to the user-defined restrictions. At the current moment, the simulator developed by Mera is able to perform optimization of up to the 63-ary OF. The following pictures illustrate some features and abilities of MERA s simulator. The first picture shows the hexagonal network layout as an example. MERA s simulator gives us the possibility to simulate: the network consists of 19 three-sector BSs (the central BS, 6 BSs of the first layer and 12 BSs of the second layer); three parameters (azimuth and downtitle of antenna beam and transmitter power) of the central BS and 6 BSs of the first layer can be varied during the optimization process; BSs of the second layer (12 BSs) are included in the network for the edge effect elimination by simulation of interference power received central BS and 6 BSs of the first layer; parameters of the second layer BSs are fixed and cannot be varied during the optimization process.
The second picture shows the outage for homogeneous network (on the left), for initial state (center) and after the optimization process (on the right). The following picture shows the SNIR for homogeneous network (on the left), for initial state (center) and after the optimization process (on the right). We propose that the transmitter power in sector 1SE is equal to zero (cell outage) and adjust 15 parameters simultaneously: azimuth direction and downtitle of the antenna beam in five sectors 1SW, 2SW, 7SW, 6N, 7N and transmitter power in these sectors. It is proposed that cell loading is full.