Characterizing the Impact of Feedback Delays on Wideband Rate Adaptation

In contemporary orthogonal frequency division multiplexing (OFDM) systems, such as Long Term Evolution (LTE), LTE-Advanced, and WiMAX, a codeword is transmitted over a group of subcarriers. Since different subcarriers see different channel gains in frequency-selective channels, the modulation and coding scheme (MCS) of the codeword must be selected based on the vector of signal-to-noise-ratios (SNRs) of these subcarriers. Exponential effective SNR mapping (EESM) maps the vector of SNRs into an equivalent flat-fading SNR, and is widely used to simplify this problem. We develop a new analytical framework to characterize the throughput of EESM-based rate adaptation in such wideband channels in the presence of feedback delays. We derive a novel accurate approximation for the throughput as a function of feedback delay. We also propose a novel bivariate gamma distribution to model the time evolution of EESM between the times of estimation and data transmission, which facilitates the analysis. These are then generalized to a multi-cell, multi-user scenario with various frequency-domain schedulers. Unlike prior work, most of which is simulation-based, our framework encompasses both correlated and independent subcarriers and various multiple antenna diversity modes; it is accurate over a wide range of delays.

Discrete-Rate Adaptation and Selection in Energy Harvesting Wireless Systems

In a system with energy harvesting (EH) nodes, the design focus shifts from minimizing energy consumption by infrequently transmitting less information to making the best use of available energy to efficiently deliver data while adhering to the fundamental energy neutrality constraint. We address the problem of maximizing the throughput of a system consisting of rate-adaptive EH nodes that transmit to a destination. Unlike related literature, we focus on the practically important discrete-rate adaptation model. First, for a single EH node, we propose a discrete-rate adaptation rule and prove its optimality for a general class of stationary and ergodic EH and fading processes. We then study a general system with multiple EH nodes in which one is opportunistically selected to transmit. We first derive a novel and throughput-optimal joint selection and rate adaptation rule (TOJSRA) when the nodes are subject to a weaker average power constraint. We then propose a novel rule for a multi-EH node system that is based on TOJSRA, and we prove its optimality for stationary and ergodic EH and fading processes. We also model the various energy overheads of the EH nodes and characterize their effect on the adaptation policy and the system throughput.

ADAPTATION OF MW LEVEL NATURAL GAS LEAN BURN ENGINE FOR PRODUCER GAS OPERATION A GRID CONNECTED POWER PLANT

Availability of producer gas engines at MW being limited necessitates to adapt engine from natural gas operation. The present work focus on the development of necessary kit for adapting a 12 cylinder lean burn turbo-charged natural gas engine rated at 900 kWe (Waukesha make VHP5904LTD) to operate on producer and set up an appropriate capacity biomass gasification system for grid linked power generation in Thailand. The overall plant configuration had fuel processing, drying, reactor, cooling and cleaning system, water treatment, engine generator and power evacuation. The overall project is designed for evacuation of 1.5 MWe power to the state grid and had 2 gasification system with the above configuration and 3 engines. Two gasification system each designed for about 1100 kg/hr of woody biomass was connected to the engine using a producer gas carburetor for the necessary Air to fuel ratio control. In the use of PG to fuel IC engines, it has been recognized that the engine response will differ as compared to the response with conventional fueled operation due to the differences in the thermo-physical properties of PG. On fuelling a conventional engine with PG, power de-rating can be expected due to the lower calorific value (LCV), lower adiabatic flame temperature (AFT) and the lower than unity product to reactant more ratio. Further the A/F ratio for producer gas is about 1/10th that of natural gas and requires a different carburetor for engine operation. The research involved in developing a carburetor for varying load conditions. The patented carburetor is based on area ratio control, consisting of a zero pressure regulator and a separate gas and air line along with a mixing zone. The 95 litre engine at 1000 rpm has an electrical efficiency of 33.5 % with a heat input of 2.62 MW. Each engine had two carburetors designed for producer gas flow each capable of handling about 1200 m3/hr in order to provide similar engine heat input at a lower conversion efficiency. Cold flow studies simulating the engine carburetion system results showed that the A/F was maintained in the range of 1.3 +/- 0.1 over the entire flow range. Initially, the gasification system was tested using woody biomass and the gas composition was found to be CO 15 +/- 1.5 % H-2 22 +/- 2% CH4 2.2 +/- 0.5 CO2 11.25 +/- 1.4 % and rest N-2, with the calorific value in the range of 5.0 MJ/kg. After initial trials on the engine to fine tune the control system and adjust various engine operating parameter a peak load of 800 kWe was achieved, while a stable operating conditions was found to be at 750 kWe which is nearly 85 % of the natural gas rating. The specific fuel consumption was found to be 0.9 kg of biomass per kWh.

Multi-scale vulnerability assessment for adaptation planning

Vulnerability of communities and natural ecosystems, to potential impacts of climate change in developing countries like India, and the need for adaptation are rapidly emerging as central issues in the debate around policy responses to climate change. The present study presents an approach to identify and prioritize the most vulnerable districts, villages and households in Karnataka State, through a multi-scale assessment of inherent vulnerability to current climate variability. It also identifies the drivers of inherent vulnerability, thereby providing a tool for developing and mainstreaming adaptation strategies, in ongoing developmental or dedicated adaptation programmes. The multi-scale assessment was made for all 30 districts at the state level in Karnataka, about 1220 villages in Chikballapur district, and at the household level for two villages - Gundlapalli and Saddapalli - in Bagepalli taluk of Chikballapur district. At the district, village and household levels, low levels of education and skills are the dominant factors contributing to vulnerability. At the village and household level, the lack of income diversification and livelihood support institutions are key drivers of vulnerability. The approach of multi-scale vulnerability assessment facilitates identification and prioritization of the drivers of vulnerability at different scales, to focus adaptation interventions to address these drivers.

ON THE UTILITY OF CANONICAL CORRELATION ANALYSIS FOR DOMAIN ADAPTATION IN MULTI-VIEW HEADPOSE ESTIMATION

The utility of canonical correlation analysis (CCA) for domain adaptation (DA) in the context of multi-view head pose estimation is examined in this work. We consider the three problems studied in 1], where different DA approaches are explored to transfer head pose-related knowledge from an extensively labeled source dataset to a sparsely labeled target set, whose attributes are vastly different from the source. CCA is found to benefit DA for all the three problems, and the use of a covariance profile-based diagonality score (DS) also improves classification performance with respect to a nearest neighbor (NN) classifier.

Discriminative language model adaptation for Mandarin broadcast speech transcription and translation

This paper investigates unsupervised test-time adaptation of language models (LM) using discriminative methods for a Mandarin broadcast speech transcription and translation task. A standard approach to adapt interpolated language models to is to optimize the component weights by minimizing the perplexity on supervision data. This is a widely made approximation for language modeling in automatic speech recognition (ASR) systems. For speech translation tasks, it is unclear whether a strong correlation still exists between perplexity and various forms of error cost functions in recognition and translation stages. The proposed minimum Bayes risk (MBR) based approach provides a flexible framework for unsupervised LM adaptation. It generalizes to a variety of forms of recognition and translation error metrics. LM adaptation is performed at the audio document level using either the character error rate (CER), or translation edit rate (TER) as the cost function. An efficient parameter estimation scheme using the extended Baum-Welch (EBW) algorithm is proposed. Experimental results on a state-of-the-art speech recognition and translation system are presented. The MBR adapted language models gave the best recognition and translation performance and reduced the TER score by up to 0.54% absolute. © 2007 IEEE.

Language model combination and adaptation using weighted finite state transducers

In speech recognition systems language model (LMs) are often constructed by training and combining multiple n-gram models. They can be either used to represent different genres or tasks found in diverse text sources, or capture stochastic properties of different linguistic symbol sequences, for example, syllables and words. Unsupervised LM adaptation may also be used to further improve robustness to varying styles or tasks. When using these techniques, extensive software changes are often required. In this paper an alternative and more general approach based on weighted finite state transducers (WFSTs) is investigated for LM combination and adaptation. As it is entirely based on well-defined WFST operations, minimum change to decoding tools is needed. A wide range of LM combination configurations can be flexibly supported. An efficient on-the-fly WFST decoding algorithm is also proposed. Significant error rate gains of 7.3% relative were obtained on a state-of-the-art broadcast audio recognition task using a history dependently adapted multi-level LM modelling both syllable and word sequences. ©2010 IEEE.

Language model cross adaptation for LVCSR system combination

State-of-the-art large vocabulary continuous speech recognition (LVCSR) systems often combine outputs from multiple subsystems developed at different sites. Cross system adaptation can be used as an alternative to direct hypothesis level combination schemes such as ROVER. In normal cross adaptation it is assumed that useful diversity among systems exists only at acoustic level. However, complimentary features among complex LVCSR systems also manifest themselves in other layers of modelling hierarchy, e.g., subword and word level. It is thus interesting to also cross adapt language models (LM) to capture them. In this paper cross adaptation of multi-level LMs modelling both syllable and word sequences was investigated to improve LVCSR system combination. Significant error rate gains up to 6.7% rel. were obtained over ROVER and acoustic model only cross adaptation when combining 13 Chinese LVCSR subsystems used in the 2010 DARPA GALE evaluation. © 2010 ISCA.