Power Consumption Trade-off Analysis in an OFDM Communication Chain
MetadataVis full innførsel
Focus has been placed on the actual power consumption of different parts in the communication system in order to establish a trade-off between each of the units and how it affects the power consumption in other units.Taking Digital Signal Processing (DSP) circuitry power consumption into account, it has previously been shown to be important for shorter communication distances where the transmission power is not dominant as traditionally believed in the total power budget. Texas instrument TMS320VC5510 that was intended for applications that require a combination of strong DSP performance and high-energy efficiency was used as our Digital Signal Processing circuitry. This circuitry primarily consists of data path, which contains four main execution units: a 40-bit ALU, 40-bit barrel shifter and two 17 x 17-bit MAC units. As power consumption can vary on this DSP device, a spreadsheet that was developed by Texas Instrument was used to provide a better estimate for the power consumption measurements. A theoretical investigation of the power consumption in the different parts of the system was carried out. Quadratic Amplitude Modulation (QAM) was the choice of the modulation scheme used in this thesis work because it uses large number of symbols and enables faster data compared to other modulation scheme. This is also the requirement for the Orthogonal Frequency Division Multiplexing (OFDM) as a multicarrier modulation scheme with larger number of subcarriers at low symbol rate. An FFT operation, which ensures that the subcarriers do not interfere with each other with different sizes, was used in the experiments. Power consumption was also calculated. It was shown that varying the clock frequency, using different constellation sizes and different FFT sizes have significant effects on the power consumption of the circuitry. By implementation, low power consumption was achieved when symbols were transmitted at larger constellation size, lower FFT size and lower clock frequency.