Adem Coskun

Blind Multidimensional Matched Filtering Techniques for Single Input Multiple Output Communications

In this paper, blind multidimensional matched filtering techniques for Single Input Multiple Output (SIMO) communications are examined. To improve the signal-to-noise ratio prior to the equalization process, three different techniques are proposed to blindly implement the multidimensional matched filtering. Different from the blind channel identification techniques for SIMO channels in the literature, the proposed techniques require neither the implementation inefficient matrix decomposition methods nor the higher-order statistics of the received data. Therefore, they are favorable over existing methods due to their simplicity in application. The matched filter equivalents are established through the use of an adaptive filter as well as the channel equalization being performed blindly. It has been shown that the equalization performance of the proposed methods are close to the matched filter bound, and to support this, a comparison between the matched filter estimation error performances of our proposed techniques is also provided. A short discussion to improve the convergence speed of the proposed approaches is also included in this paper.

Transmit Precoding for Flat-Fading MIMO Multiuser Systems With Maximum Ratio Combining Receivers

We examine the application of transmit precoding in multiuser multi-input-multi-output (MIMO) communication systems with maximum ratio combining (MRC) receivers. In many multiuser applications, the maximum-likelihood or minimum mean-square error (MMSE) receivers can be prohibitive to implement due to their high implementation complexity. We examine the performance of the system with simple MRC receivers and carefully selected precoders, which are designed to compensate the lack of high-complexity receivers, at the transmitter side. We examine the sum MSE minimization and signal-to-interference-plus-noise ratio (SINR) balancing frameworks for the selection of precoders. The performance of two frameworks with MRC receivers are compared between themselves as well as with their counterparts implementing MMSE receivers. It has been observed that the SINR balancing framework with simple MRC receivers has little performance loss in comparison with the MMSE receivers with a proper selection of precoders.

All-Adaptive Blind Matched Filtering for the Equalization and Identification of Multipath Channels—A Practical Approach

Blind matched filter receiver is advantageous over the state-of-the-art blind schemes due the simplicity in its implementation. To estimate the multipath communication channels, it uses neither any matrix decomposition methods nor statistics of the received data higher than the second order ones. On the other hand, the realization of the conventional blind matched filter receiver requires the noise variance to be estimated and the equalizer parameters to be calculated in state-space with relatively costly matrix operations. In this paper, a novel architecture is proposed to simplify a potential hardware implementation of the blind matched filter receiver. Our novel approach transforms the blind matched filter receiver into an all-adaptive format which replaces all the matrix operations. Furthermore, the novel design does not need for any extra step to estimate the noise variance. In this paper we also report on a comparative channel equalization and channel identification scenario, looking into the performances of the conventional and our novel all-adaptive blind matched filter receiver through simulations.

A blind implementation of multi-dimensional matched filtering in a Maximum-Likelihood receiver for SIMO channels

In order to establish the optimal receiver strategy, in terms of error rate for Single Input Multi Output (SIMO) wireless channels, the Maximum Likelihood (ML) detection should be performed following a multi-dimensional matched filter. However, the implementation of the matched filter and the ML detection both need the estimation of the channel impulse response in advance. In this work, we propose a novel method to establish the matched filters of the SIMO channel blindly alongside a three-step technique for the blind and adaptive ML detection of the symbol vector. With the use of the novel method, the system will benefit from the bandwidth efficiency point of view due to the use of blind schemes. The constant modulus algorithm is utilized to perform the blind matched filtering operation and later Least Mean Squared algorithm is introduced for further correction on the matched filter estimate. The blindly estimated matched filters are incorporated into the ML detector so that the transmitted symbols are found and therefore the channel is equalized. Simulations are provided to present the equalization performance and convergence speed of the novel technique.

Low complexity all-pass based polyphase decimation filters for ECG monitoring

This paper presents a low complexity high efficiency decimation filter which can be employed in EletroCardioGram (ECG) acquisition systems. The decimation filter with a decimation ratio of 128 works along with a third order sigma delta modulator. It is designed in four stages to reduce cost and power consumption. The work reported here provides an efficient approach for the decimation process for high resolution biomedical data conversion applications by employing low complexity two-path all-pass based decimation filters. The performance of the proposed decimation chain was validated by using the MIT-BIH arrhythmia database and comparative simulations were conducted with the state of the art.

The design of low complexity low power pipelined short length Winograd Fourier transforms

In this paper a novel pipelining approach applicable to Winograd Fourier transforms is presented. The novel approach makes use of reconfigurable multiplier blocks to implement the real multipliers required for the transform as well as sharing the hardware resources among additions. The additions are realized using modified forms of butterfly circuits. The novel approach is tested on a 5-point Winograd Fourier transform and the circuit area and power dissipation of the design are estimated using an in-house power estimation tool and compared to the state-of-the-rt approaches.

Blind correlation-based DFE receiver for the equalization of single input multi output communication channels

In this paper, the correlation-based Decision Feedback Equalizer (DFE), where the received data from multiple antennas are processed by a multi-dimensional matched filter and then combined prior to the equalization with a single input single output DFE, is discussed and its blind implementation is introduced. To perform the correlation-based DFE blindly, the multi-dimensional matched filter is replaced by an adaptive filter and the DFE filter weights are calculated via manipulating over the second order statistics of the received data. In the blind architecture, the adaptive filter converges to matched filter equivalents, therefore the matched filters of the corresponding communication channels are also blindly be estimated in addition to the blind equalization process. The mean-squared error of the estimation of matched filters and the equalization performance of the proposed blind architecture are also studied and simulated.

Testing and Analysis of Activities of Daily Living Data with Machine Learning Algorithms

It is estimated that 28% of European Union’s population will be aged 65 or older by 2060. Europe is getting older and this has a high impact on the estimated cost to be spent for older people. This is because, compared to the younger generation, older people are more at risk to have/face cognitive impairment, frailty and social exclusion, which could have negative effects on their lives as well as the economy of the European Union. The ‘active and independent ageing’ concept aims to support older people to live active and independent life in their preferred location and this goal can be fully achieved by understanding the older people (i.e their needs, abilities, preferences, difficulties they are facing during the day). One of the most reliable resources for such information is the Activities of Daily Living (ADL), which gives essential information about people’s lives. Understanding this kind of information is an important step towards providing the right support, facilities and care for the older population. In the literature, there is a lack of study that evaluates the performance of Machine Learning algorithms towards understanding the ADL data. This work aims to test and analyze the performance of the well known Machine Learning algorithms with ADL data.

A wireless sensor network for early forest fire detection and monitoring as a decision factor in the context of a complex integrated emergency response system

Wildfires, often dubbed megafires, have in recent times increased in both frequency and scale, owing largely to human error as well as climate change. Due to their uncontrolled unpredictable rapid growth and behaviour, they can quickly become difficult to contain, leading to significant loss of lives, wildlife and property. It is therefore critical to tackle such fires in the early stages. This demands a reduction in the initial time to detection while ensuring reliability and a low false alarm rate. This paper discusses a new category of compact, easily deployable and energy efficient approach to sensor nodes for the continued monitoring of forest environments as well as the early detection of fires in their infancy based on a combination of sensory inputs. The sensor network reported in this paper was tested with other subsystems/technologies, in a real-life firefighting trial as part of a coordinated firefighting scenario with promising results.

Multiplier Free Implementation of 8-Tap Daubechies Wavelet Filters for Biomedical Applications

Due to an increasing demand for on-sensor biosignal processing in wireless ambulatory applications, it is crucial to reduce the power consumption and hardware cost of the signal processing units. Discrete Wavelet Transform (DWT) is very popular tool in artifact removal, detection and compression for time-frequency analysis of biosignals and can be implemented as two-branch filter bank. This work proposes a new, completely multiplier free filter architecture for implementing Daubechies wavelets which targets Field-Programmable-Gate-Array (FPGA) technologies by replacing multipliers with Reconfigurable Multiplier Blocks (ReMBs). The results have shown that the proposed technique reduces the hardware complexity by 25 % in terms of Look-up Table (LUT) count and can be used in low-cost embedded platforms for ambulatory physiological signal monitoring and analysis.