Article Preview
TopIntroduction
The ever-increasing demand for wireless access is in constant competition with the inherently finite radio spectrum and limited allocations for wireless services. Increasing the utility of the existing spectrum requires that new technologies efficiently use allocated spectrum or operate as non-interfering secondary users in existing spectrum. Personal area networks (PANs) contain examples of such technologies, including ZigBee (Wheeler, 2007), Bluetooth (Haartsen, 2000), and others, which share access to the 2.4 GHz band. These systems usually exhibit modest bitrates and short ranges appropriate to their application.
For applications requiring that the data source (hereafter, the tag), be both low-cost and low-power, radio frequency (RF) backscatter presents a viable method of short-range communications. Typical RF identification (RFID) tags eliminate most components of a traditional radio and instead simply modulate the impedance of the tag’s antenna (Want, 2006). This in turn modulates the fraction of an incident wave reflected by the tag. The reader (receiver) senses the reflections to determine the transmitted data. In addition, the reader typically supplies the illuminating energy to the tag, which must be of a sufficient strength to overcome the 
range losses.
Recent work explores backscatter communications using atypical sources of incident illumination. Liu, et al. (2013) recently proposed using noncoherently-detected “ambient backscatter” of television signals for communications over less than a meter. Blunt, et al. (2010) proposed “radar-embedded communications” to obscure signals in the modified backscatter of incident radar pulses. These approaches can simplify the communications system by relying on externally-provided illuminators. Additionally, these systems can benefit from signal obfuscation, since the backscatter from the tag is much weaker than both the ambient signal and reflections from the environment around the tag.
This paper extends on the concepts presented by Liu and Blunt by applying coherent passive radar processing to ambient backscatter communications. Early work on this concept was presented by Barott (2014), and this paper presents both a more-detailed analysis and new results. Key differences of the proposed approach from that given by Liu, et al. (2013) include coherent processing and the removal of direct path interference (DPI). These steps enable a significantly increased range, and the link budget suggests that data rates of 103 bps might be achievable at ranges of 1 km. The cost of this approach is an increased computational complexity of the receiver, but receiver costs will decrease with future improvements in computing capabilities.
This paper proceeds according to the following outline. First, descriptions of the theory of the proposed approach and relevant passive processing algorithms are presented. This is followed by a description of some practical characteristics that are unique to the proposed technique, including geometric limitations imposed by the bistatic geometry. Next, results from proof-of-concept experiments are described, which are followed by discussion, analysis, and conclusions.