Design and Implementation of Software Defined Radios on a Homogeneous Multi-Processor Architecture
Airoldi, Roberto (2013)
Airoldi, Roberto
Tampere University of Technology
2013
Teknis-taloudellinen tiedekunta - Faculty of Business and Technology Management
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:ISBN:978-952-15-3096-8
https://urn.fi/URN:ISBN:978-952-15-3096-8
Tiivistelmä
In the wireless communications domain, multi-mode and multi-standard platforms are becoming increasingly the central focus of system architects. In fact, mobile terminal users require more and more mobility and throughput, pushing towards a fully integrated radio system able to support different communication protocols running concurrently on the platform. A new concept of radio system was introduced to meet the users' expectations. Flexible radio platforms have became an indispensable requirement to meet the expectations of the users today and in the future.
This thesis deals with issues related to the design of flexible radio platforms. In particular, the flexibility of the radio system is achieved through the concept of software defined radios (SDRs). The research work focuses on the utilization of homogeneous multi-processor (MP) architectures as a feasible way to efficiently implement SDR platforms. In fact, platforms based on MP architectures are able to deliver high performance together with a high degree of flexibility. Moreover, homogeneous MP platforms are able to reduce design and verification costs as well as provide a high scalability in terms of software and hardware. However, homogeneous MP architectures provide less computational efficiency when compared to heterogeneous solutions.
This thesis can be divided into two parts: the first part is related to the implementation of a reference platform while the second part of the thesis introduces the design and implementation of flexible, high performance, power and energy efficient algorithms for wireless communications. The proposed reference platform, Ninesilica, is a homogeneous MP architecture composed of a 3x3 mesh of processing nodes (PNs), interconnected by a hierarchical Network-on-Chip (NoC). Each PN hosts as Processing Element (PE) a processor core. To improve the computational efficiency of the platform, different power and energy saving techniques have been investigated. In the design, implementation and mapping of the algorithms, the following constraints were considered: energy and power efficiency, high scalability of the platform, portability of the solutions across similar platforms, and parallelization efficiency.
Ninesilica architecture together with the proposed algorithm implementations showed that homogeneous MP architectures are highly scalable platforms, both in terms of hardware and software. Furthermore, Ninesilica architecture demonstrated that homogeneous MPs are able to achieve high parallelization efficiency as well as high energy and power savings, meeting the requirements of SDRs as well as enabling cognitive radios.
Ninesilica can be utilized as a stand-alone block or as an elementary building block to realize clustered many-core architectures. Moreover, the obtained results, in terms of parallelization efficiency as well as power and energy efficiency are independent of the type of PE utilized, ensuring the portability of the results to similar architectures based on a different type of processing element.
This thesis deals with issues related to the design of flexible radio platforms. In particular, the flexibility of the radio system is achieved through the concept of software defined radios (SDRs). The research work focuses on the utilization of homogeneous multi-processor (MP) architectures as a feasible way to efficiently implement SDR platforms. In fact, platforms based on MP architectures are able to deliver high performance together with a high degree of flexibility. Moreover, homogeneous MP platforms are able to reduce design and verification costs as well as provide a high scalability in terms of software and hardware. However, homogeneous MP architectures provide less computational efficiency when compared to heterogeneous solutions.
This thesis can be divided into two parts: the first part is related to the implementation of a reference platform while the second part of the thesis introduces the design and implementation of flexible, high performance, power and energy efficient algorithms for wireless communications. The proposed reference platform, Ninesilica, is a homogeneous MP architecture composed of a 3x3 mesh of processing nodes (PNs), interconnected by a hierarchical Network-on-Chip (NoC). Each PN hosts as Processing Element (PE) a processor core. To improve the computational efficiency of the platform, different power and energy saving techniques have been investigated. In the design, implementation and mapping of the algorithms, the following constraints were considered: energy and power efficiency, high scalability of the platform, portability of the solutions across similar platforms, and parallelization efficiency.
Ninesilica architecture together with the proposed algorithm implementations showed that homogeneous MP architectures are highly scalable platforms, both in terms of hardware and software. Furthermore, Ninesilica architecture demonstrated that homogeneous MPs are able to achieve high parallelization efficiency as well as high energy and power savings, meeting the requirements of SDRs as well as enabling cognitive radios.
Ninesilica can be utilized as a stand-alone block or as an elementary building block to realize clustered many-core architectures. Moreover, the obtained results, in terms of parallelization efficiency as well as power and energy efficiency are independent of the type of PE utilized, ensuring the portability of the results to similar architectures based on a different type of processing element.
Kokoelmat
- Väitöskirjat [4862]