Vesa Hirvisalo

A Reconfigurable Multi-standard Radio Platform

This paper presents a dynamically reconfigurable multiradio RF architecture concept, which can be used for RF platform and control optimization. The platform realization is based on the RF hardware and its configuration mechanisms. The related control software is realized through functional separation of configuration management and timing control. Both key HW and SW elements are discussed and optimization opportunities evaluated using high level analysis on key building blocks.

Latency and throughput characterization of convolutional neural networks for mobile computer vision

We study performance characteristics of convolutional neural networks (CNN) for mobile computer vision systems. CNNs have proven to be a powerful and efficient approach to implement such systems. However, the system performance depends largely on the utilization of hardware accelerators, which are able to speed up the execution of the underlying mathematical operations tremendously through massive parallelism. Our contribution is performance characterization of multiple CNN-based models for object recognition and detection with several different hardware platforms and software frameworks, using both local (on-device) and remote (network-side server) computation. The measurements are conducted using real workloads and real processing platforms. On the platform side, we concentrate especially on TensorFlow and TensorRT. Our measurements include embedded processors found on mobile devices and high-performance processors that can be used on the network side of mobile systems. We show that there exists significant latency-throughput trade-offs but the behavior is very complex. We demonstrate and discuss several factors that affect the performance and yield this complex behavior.

Elastic GPU processing for streaming video data

We study using multiple GPU devices to accelerate processing of streaming data. The amount of traffic on the Internet is rapidly rising. Much of this rise is due to media streams that need to be processed. However, the global growth rate of processing capacity is lagging behind. We see Fog computing, i.e., processing capacity attached directly to the nodes of the network, as one possibility to address this problem. In this paper, we present our approach to attach scalable processing capacity to network nodes. We use multiple generic purpose accelerators. To show the viability of our approach, we present our simulations and measurements related to a setup where multiple GPUs are attached to an NPU to accelerate processing of video streams. We show that GPU virtualization and communication batching can be used to get scalable processing.

Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering

This paper presents a dynamically reconfigurable multiradio RF architecture concept, which can be used for RF platform and control optimization. The platform realization is based on the RF hardware and its configuration mechanisms. The related control software is realized through functional separation of configuration management and timing control. Both key HW and SW elements are discussed and optimization opportunities evaluated using high level analysis on key building blocks.