Object detection and classification are the basic tasks in video analytics and become the starting point for other complex applications. Traditional video analytics approaches are manual and time consuming. These are subjective due to the very involvement of human factor. We present a cloud based video analytics framework for scalable and robust analysis of video streams. The framework empowers an operator by automating the object detection and classification process from recorded video streams. An operator only specifies an analysis criteria and duration of video streams to analyse. The streams are then fetched from a cloud storage, decoded and analysed on the cloud. The framework executes compute intensive parts of the analysis to GPU powered servers in the cloud. Vehicle and face detection are presented as two case studies for evaluating the framework, with one month of data and a 15 node cloud. The framework reliably performed object detection and classification on the data, comprising of 21,600 video streams and 175 GB in size, in 6.52 hours. The GPU enabled deployment of the framework took 3 hours to perform analysis on the same number of video streams, thus making it at least twice as fast than the cloud deployment without GPUs.

No abstract

The Channel Matched Filter Decision Feedback Equaliser (CMF-DFE) is a high performance equalisation method with reduced computational complexity; which is essential for high-speed communication systems with severe intersymbol interference. The method exploits the fact that matched filtering of a wideband channel results in a symmetrical channel profile centred on a realvalued peak while exploiting multipath diversity. This paper describes the implementation of a HIPERLAN/l compatible equaliser using the CMF-DFE method. The performance of the implemented algorithm and the implementation benchmarks are given for the Hiperladl standard. Results are also given for a number of different modulation schemes.

No abstract

Coded Orthogonal Frequency Division Multiplexing (COFDM) is currently specified in all three of the world's 5 GHz wireless LAN standards (Hiperlan/2, IEEE 802.11a and MMAC HiSWANa). This technology was chosen due to its robustness at high data rates in a frequency selective multipath channel. Each standard operates using adaptive sub-band Quadrate Amplitude Modulation (QAM) and offers a maximum data rate of 54 Mbits/s over a 20 MHz channel. This paper describes a real-time DSP implementation of an asynchronous OFDM based high speed WLAN system. The software reconfigurable OFDM based platform is developed around the Texas Instruments fixed point TMS320C6201 DSP. The physical layer DSP performance is evaluated and compared for an indoor channel against floating point C++ based simulations. Data throughput and complexity estimates are generated from the resulting hardware platform. Finally, a video based communications application is developed to operate over the demonstrator. Results indicate that the fixed point DSP solution can operate within 0.5 dB of the floating point simulation in an AWGN channel. For the indoor fading channel, an implementation loss of around 2.5 dB was observed.