Pilot Tones Design for Channel Estimation Using Elephant Herding Optimization Algorithm in Massive MIMO Systems

Today, the increase in the demand for mobile communication and the increasing need for data transfer have reached great dimensions. In order to meet this need, multi-input multi-output (MIMO) can be increased by significantly increasing the number of antennas in the base station by making more use of the spatial multiplexing capability. With the significant increase in the number of antennas, the concept of massive MIMO has emerged. In massive MIMO systems, like many other communication systems, the channel status information of the channels must be obtained. Therefore, channel estimation methods are used to meet the need for channel state information. The least squares algorithm, which is one of the simple channel estimation techniques, is one of the most preferred techniques in this field. In this paper, pilot tone optimization was applied in the least squares channel estimation method by using elephant herding optimization (EHO) technique in massive MIMO systems. When performance of EHO is compared with performances of genetic algorithm, particle swarm optimizations, invasive weed optimization, harmony search and random forest algorithm it is seen that the EHO is the most successful algorithm. For example, in the calculations made in cases where the signal-to-noise ratio value is 18 dB, mean squared error was calculated as 2.88 × 10 –5 with genetic algorithm, 2.76 × 10 –5 with particle swarm optimization, 2.75 × 10 –5 with invasive weed optimization, 2.60 × 10 –5 with harmony search, and 2.63 × 10 –5 with random forest algorithm, while it was calculated as 2.49 × 10 –5 with EHO. When the pilots' positions were determined using EHO, the number of erroneous bits sent significantly decreased compared to both random placement and placement at equal intervals. This situation can be considered a significant achievement for a channel estimation process that involves extremely low processing load.