Rafael Cepeda

Privacy for Smart Meters: Towards Undetectable Appliance Load Signatures

Smart grid privacy encompasses the privacy of information extracted by analysing smart metering data. In this paper, we suggest that home electrical power routing can be used to moderate the homes load signature in order to hide appliance usage information. In particular, 1) we introduce a power management model using a rechargeable battery, 2) we propose a power mixing algorithm, and 3) we evaluate its protection level by proposing three different privacy metrics: an information theoretic (relative entropy), a clustering classification, and a correlation/regression one; these are tested on different metering datasets. This paper sets the ground for further research on the subject of optimising home energy management with regards to hiding load signatures.

Long-term measurements of spectrum occupancy characteristics

This paper describes the results from a series of long-term observations of spectrum occupancy in the range 300 MHz–4.9 GHz made at a single location. Over 6 months of data has been gathered from a measurement system that is designed to operate continuously, allowing a full picture of spectrum occupancy to be built up. The large amount of data captured permits analysis to identify the portions of spectrum that are occupied non-continuously (i.e. with a duty cycle of less than 100%); these channels might be most suitable for use by a cognitive radio (CR) system that is able to identify temporary spectrum holes and then exploit them. The aim of performing a comprehensive analysis of channel occupancy and its variability is to improve the potential for smart spectrum access by CR devices, by informing their choices concerning the particular portions of spectrum to scan, and how frequently. Results allow measurement channels to be categorised according to their duty cycle; spectrum with a duty cycle between 10% and 90% is considered to be particularly suitable and is found in the regions below 500 MHz and between 1 GHz and 1.2 GHz. A predictable pattern of time occupancy is discovered, caused by variable activity in the portions of spectrum allocated to cellular systems. The characteristics of the occupancy duty cycle according to the time of day is also investigated; some frequencies are shown to exhibit considerable variation of occupancy depending on the hour of observation. A subset of channels is selected for more detailed investigation including short-term variations in channel occupancy.

ElecPrivacy: Evaluating the Privacy Protection of Electricity Management Algorithms

The data collected by a home smart meter can potentially reveal sensitive private information about the home resident(s). In this paper, we study how home energy resources can be used to protect the privacy of the collected data. In particular we: a) introduce a power mixing algorithm to selectively protect a set of consumption events; b) develop a range of different privacy protection metrics; c) analyze real smart metering data sampled twice a minute over a period of 13 days; and d) evaluate the protection offered by different power mixing algorithms. Major factors which determine the efficiency of the proposed power mixing algorithms are identified, such as battery capacity and power, and user preferences for privacy-based allocations of battery energy quotas.

Privacy protection system and metrics for hiding electrical events

Smart grid privacy concerns the privacy of information extracted by analysing smart metering data. We present ElecPrivacy: a home electrical power management system that uses a rechargeable battery to mask home energy load signatures and, effectively, protect the privacy of appliance usage information. ElecPrivacy can be studied in the context of the classic communications problem, where input data is passed through a communication channel that distorts it. In this paper, we define and measure how the appearance of ElecPrivacy events can be estimated, or, reversely, how well the secrecy of this data is protected. In particular, we develop a range of privacy metrics by combining clustering, information theoretic (K-divergence), correlation and regression techniques, and testing over a large data set obtained from real home measurements.

An Improved Antenna Mounting for Ultra-Wideband On-Body Communications and Channel Characterization

The electromagnetic coupling of antennas with the human body is a well-known, but often neglected, problem in body area network (BAN) communications. The resulting distortion of the antennas free-space characteristics may, in turn, demand more complex terminals to cope with the different conditions encountered as antennas are mounted on different parts of the body and/or varying separations from the skin. Therefore, in this paper, we propose a mounting method to provide a more stable response and to reduce the antennas coupling to the body for ultra-wideband (UWB) BAN applications. The proposed mounting method is evaluated on two UWB BAN antennas, characterizing the antennas coupling with the body by analyzing variations in the antennas reflection coefficient between 4-9 GHz, when mounted at different distances from the skin and at different locations on the body. The modified mounting method was found to increase the correlation coefficient between the antennas reflection coefficient when in free space and mounted using 5 mm of spacer, for the two antennas, from 0.22 and 0.13, to 0.66 and 0.90. The free-space characteristics of the modified antennas are also tested, showing a reduction of up to 4.7 dB of radiated power into the body.

The Measurement of Frequency Dependent Path Loss in Residential LOS Environments using Time Domain UWB Channel Sounding

Future ultra wide-band (UWB) systems will probably use bandwidths exceeding 1 GHz to achieve the data rates necessary for wireless video streaming and rapid file transfer. It is therefore essential to consider how propagation losses vary as a function of frequency, especially because the dynamic range of commercial UWB receivers is limited by the analogue-to-digital converter (ADC). This paper characterizes the line-of-sight (LOS) frequency dependent path losses encountered in a residential environment using a prototype UWB time domain channel sounder. Importantly, results show that the frequency dependent path loss exponent varies as much over small areas (0.1 m2), as it does over widely spaced areas.

MIMO-OFDM Pilot Placement Algorithms for Wideband Indoor Communications

To facilitate coherent detection of orthogonal frequency division multiplexed transmissions, pilot symbols can be transmitted in some of the subcarriers. Their placement to minimise the mean squared error in channel estimation is considered. Such problems have been widely addressed previously for channels subject to wide sense stationary uncorrelated scattering processes. While the stationarity assumption is usually realistic, indoor wideband channels, for example, are subject to significantly correlated scattering processes. Furthermore, practical multiple antenna transmitters/receivers are bound to exhibit spatial correlations. This work considers pilot placement optimisation in a multiple antenna setting subject to arbitrary correlations in the delay and spatial domains. Noting that the naive approach incurs a prohibitive complexity, low complexity greedy solutions are developed. It is further proven that the conventional equi-spaced placement becomes optimal at high SNR even with correlations, when the correlation matrices are of full rank. A measurement campaign is conducted to estimate real-world channel statistics in an indoor wideband environment, and observe the applicability of developed algorithms. It is confirmed that practical channels are correlated to the extent of exhibiting rank deficiencies, and that the proposed placement algorithms result in improved channel estimation performance.

On the measurement and simulations of the frequency dependent path loss and MB-OFDM

With the imminent deployment of ultra wideband (UWB) wireless transmission systems, it becomes fundamental to analyse and predict possible drawbacks of current technologies with the help of propagation measurements. Now that multi-band orthogonal frequency division multiplexing (MB-OFDM) is preferred for cable replacement, wireless universal serial bus (USB) - version 2.0 - in particular, its analysis is of critical interest. This paper presents measurements of the frequency dependent path loss for an indoor residential location, and then shows the performance of MB-OFDM under the worst case scenario conditions. Results show that for a frequency hopped transmission, design alternatives are necessary to maintain the expected data throughput.

The performance of robust adaptive modulation over wireless channels with non reciprocal interference

Adapting a modems modulation scheme can significantly enhance the resulting performance over a wireless communications link. Non-adaptive modems are generally designed conservatively for worst-case channel conditions. Adaptive modems are able to exploit the higher channel capacity when good conditions are encountered. A number of different metrics can be used to drive the adaptation process (i.e. automatic repeat request (ARQ) or segment error rate (SER)). Not surprisingly, certain metrics are far more successful than others. This paper presents an investigation of adaptive modulation in the presence of interference. In particular, the case of non reciprocal interference is studied (i.e., cases where the level of interference is different on the forward and reverse links). Research results have shown that if adaptation is performed assuming a reciprocal channel, then poor performance will result for the link experiencing a higher level of interference. In this study, an algorithm is used to calculate a measure of the link quality. This value is then fed back to the other end of the link to adapt subsequent transmissions. Results indicate that the throughput of such a system can be considerably improved when transmitting adaptively using any one of DBPSK, DQPSK or 8DPSK modulation schemes. It is proposed that this simple approach be adopted for short range, ad hoc, wireless personal area networks (WPAN).

Power Loading in Parallel Diversity Channels Based on Statistical Channel Information

In this paper, we show that there exists an arbitrary number of power allocation schemes that achieve capacity in systems operating in parallel channels comprised of single-input multiple-output (SIMO) Nakagami-m fading subchannels when the number of degrees of freedom L (e.g., the number of receive antennas) tends to infinity. Statistical waterfilling - i.e., waterfilling using channel statistics rather than instantaneous channel knowledge - is one such scheme. We further prove that the convergence of statistical waterfilling to the optimal power loading scheme is at least O(1/(L log L)), whereas convergence of other schemes is at worst O(1/ log L). To validate and demonstrate the practical use of our findings, we evaluate the mutual information of example SIMO parallel channels using simulations as well as new measured ultrawideband channel data.