Understand your algorithms better
By Vikram Singh
Academia has paved a way for industry by designing state of the art algorithms and methods such as orthogonal frequency division multiplexing and access (OFDM(A)), LDPC codes, Polar codes etc. However lately, the contributions (innovations) per paper has drastically reduced in academia compared to the previous decades. One major reason behind this is a disconnect from the actual problems/challenges and industry itself. Many a times the assumptions considered for research works are outright impractical. Many good algorithms seems to be performing reasonably well link level evaluations, however, fails to replicate the same performance in system level simulation.
- Even good researcher fails to see certain aspects of the system such as limits of feedback reporting, latency constraints etc.
- The problem itself have a large span making it difficult to put a simulation setup in place.
This happens because of multiple reasons:
Both these problems can be handled easily without much efforts if a system level simulator is available. The researchers can design their algorithms and plug them into the simulator based on the simulator architecture. This allows the researcher to investigate the end-to-end performance of their algorithm(s). The performance criteria are, but not limited to:
- Complexity = number of flops/implementation ⇛ Indicates the power consumption and delays constraints of devices.
- Throughput = data-rate and/or spectral efficiency and/or energy efficiency.
- Scalability = performance variations with number of devices.
The system level simulator such as BWSim 4G|5G allows the researchers to understand their algorithms qualitatively. Moreover, such simulators are crucial for system and standards-based research.
For these purposes, Gigayasa Wireless has designed 4G and 5G system level simulators, called BWSim*-4G and BWSim-5G-R15, which are based on release-14 and release-15 of 3GPP, respectively.
*expanded as Broadband Wireless Simulator
