Ahmad Kamsani Samingan

Performance study of LTE experimental testbed using OpenAirInterface

The next generation, 5G wireless access, is going to support a wide range of new applications and use cases, with the capabilities including very high achievable data rate, very low latency, ultra-high reliability and the possibility to handle extreme device densities. To avoid the costly deployment, operation and maintenance of future mobile network, Radio Access Network (RAN) virtualization or Cloud RAN is the answer to the problem. The idea is to move the baseband processing to the data center and run the RAN L1, L2 and L3 protocol layers using the commodity hardware, such as high-performance general purpose processors. An open source software-based LTE implementation, such as OpenAirInterface (OAI), is definitely accelerating the RAN cloudization and also realizing the possibility of low cost LTE network deployment in the future. In this paper, we describe the OAI LTE implementation emphasizing on the user plane data flow. We have successfully emulated over-the-air transmission for 1 UE and 1 eNB LTE network supporting both FDD Band 5 and TDD Band 38. We have also performed a thorough profiling of OAI, in terms of execution time, on the user plane data flow. Our results could be served as the reference for future optimization by open source community.

Frequency-domain channel tracking for MIMO-OFDM systems, validated with an experimental data in 5.2 GHz wireless channel

This paper proposes a channel tracking technique for MIMO-OFDM system, in order to solve timing sensitivity issue with the pilot-based Channel Estimator (CE) reported in [1]. The proposed technique exploits the pilot symbols embedded within the OFDM symbol to estimate the MIMO channel coefficients. Then the estimated channel values of the pilot subcarriers, as well as the initial LTF-based channel estimated values are processed to provide a weighting factor, which is used to track the channel coefficients for the entire transmission packet. The performance of the proposed CE is evaluated with the aid of actual captured data. Our performance study considers a Non-LOS indoor environment, where the transmitter and receiver are approximately 15 meters apart, and located in different rooms. Our study shows that the proposed CE successfully mitigates the susceptibility of the original pilot-based CE to the timing synchronization error.

LTF-based vs. pilot-based MIMO-OFDM Channel Estimation algorithms: An experimental study in 5.2 GHz wireless channel

This paper outlines two classes of Channel Estimation (CE) techniques, namely the Long Training Field (LTF)-based CE as well as the pilot-based CE. Both LTF and pilot patterns are taken from the 802.11n Draft 2.0 standard. The performance of these two CE classes is evaluated with the aid of actual captured data, provided by experiments under two different propagation scenarios. The first scenario represents the slow time-varying channels, which involve Line-of-Sight (LOS) outdoor transmission over a distance of 7.5 km. By contrast, the second case corresponds to a Non-LOS indoor environment, where the transmitter and receiver are approximately 15 meters apart, and located in different rooms. Our experimental study shows that the LTF-based CE outperforms the pilot-based CE for slow time-varying outdoor LOS channels, while for indoor NLOS channels, it fails completely. On the other hand, pilot-based CE works in both test cases. Unfortunately, this method is very sensitive to timing synchronization.

Cell drop threshold architecture for multi-class shared buffer with finite memory size

Shared buffer is commonly used to utilize the buffer in the switch. In order to minimize the cell lost of high class traffic in multi-class switch, the threshold is set to drop the low class cells in the shared buffer. This will give more space to accommodate the high class traffic cells. In this paper, we analyse the performance of shared buffer with different threshold settings. The multi-class shared buffer architecture is developed for 16×16 ports switch, which is targeted for Xilinx FPGA implementation. The performance of the multi-class shared buffer switch is analysed in term of the achievable throughput as well as the drop probability. Based on the simulation with different threshold settings, it is observed that the optimum selection of cell drop threshold depends on the size of the shared buffer that triggers the RAM threshold.

Receiver antenna synchronization and carrier frequency offsets pre-mitigation for distributed massive MIMO system

This paper presents a receiver antenna synchronization and carrier frequency offsets pre-mitigation procedure, which is proposed for distributed massive MIMO system. The techniques performance is dependent to the SNR as well as the number of OFDM symbols per transmission packet. It has been shown that the procedure is working using a rapid prototyping platform based on WARP board. The MIMO system employing 16 receivers antenna has been shown to provide approximately 15dB transmit energy saving gain compared to a two-antenna system.

Improved link level LTE scheduler

Long Term Evolution (LTE) packet scheduler is a key mechanism in the LTE traffic handling. It is mainly in charge of allocating physical resource blocks for the attached UEs. The main problem in LTE scheduler is to maximize the throughput while maintaining the fairness. The performance of scheduler is very critical, especially when many UEs are located at the cell edge. The new proposed scheduler can give better cell throughput performance and at the same time maintain the fairness. The findings from the performance results show an increased up to 16 percent of cell throughput compared to proportional fair (PF) scheduler. The result also shows that the fairness level is almost the same as the PF scheduler.