December 2014 – January 2015 Intel Project Monthly Report
Next-Generation Cooperative Vehicle Safety System
Current team members:
Champion: Dr. Xingang Guo & Dr. Richard Roberts
PI and co-PI: Prof. Hsin-Mu (Michael) Tsai and Dr. Kate Ching-Ju Lin
Students: Hao-Min Lin (Ph.D.), Chia-Fu Li (Master), Hua-Yen Tseng (Master), Hsin-I Wu (Master),
Hui-Yu Lee (Master), Chung-Lin Chan (Master), Jing-Yeu Chen (Master), Ai-Ling Chen (Master),
You-Lin Wei (Master)
Communication with the champions:
1. In December 17, Michael, Rick, and several members of Intel Japan (Nakagawa, Yoshikatsu;
Ota, Yoshihiko) had a short teleconference to talk about the possibility of setting up a
CamCom demo at Intel Japan’s Collaborative Center. Intel Japan members briefly talked
about their current related projects and visions and Michael briefly introduced our
CamCom prototype. No follow-up discussion yet after the teleconference.
2. As part of the process to put our RS-FSK waveform in the new IEEE 802.15.7r1 standard,
Rick (who leads the IEEE efforts) requested for tutorial contribution for the upcoming Berlin
IEEE802 meeting. Michael submitted the materials for the tutorial presentation in
mid-January.
Progress:
1. Prototype of Smart Glasses with CamCom-enabled Visual Association
To demonstrate how CamCom can be used to enable visual association in smart glasses, we
developed a prototype using Epson Moverio BT-200 smart glasses combined with a PointGrey
USB camera mounted on the side of the glasses. The prototype is able to decode the
information embedded in the light, and then overlay the information on top of the object or
light transmitting that information in the screen of the smart glasses, visually associating the
received information with the transmitter. In our prototype, we have two indoor lights, one
transmitting the current time (a virtual clock) and the other transmitting a logo (Intel logo or
NTU logo). The image of either the clock or the logo is then displayed on top of the
transmitting lights. The system is demonstrated in the January’15 open house event as well as
in the face-to-face meeting.
2.
Expand the bandwidth of CamCom systems with multiple cameras:
In the CamCom protocol we purposed last year, the range of frequency used in
transmitting (i.e. the bandwidth) is constrained by the rolling shutter sampling rate of the
CMOS camera and the size of the transmitting light in the image. If the transmitted
frequency is higher than the rolling shutter sampling frequency, aliasing would be
observed and the receiver would be unable to determine the transmitted frequency
(because there are more than 1 possibilities). The figure below shows an example.
The x-axis represents the actual transmitted frequency while the y-axis represents the
observed frequency.
This problem can be solved by using more than one camera, each with different rolling
shutter sampling rate (i.e., different read-out time). The obtained signal will follow the
Nyquist sampling theorem, i.e., if the transmitted frequency is higher than half of the
sampling frequency, the observed frequency will be “folded” in a zig-zag manner and
forms a cycle whose “period” is the sampling frequency, as shown in the figure below.
As a result, we can use, for example, two cameras with different sampling rate to
determine the original transmitted frequency, which can now be higher than the sampling
rate of both cameras, shown below:
With this approach, the usable range of frequency, i.e., bandwidth, can be greatly
expanded for the CamCom system. We plan to explore this new idea in the next couple of
months.
Brief plan for next month:
1. Design MAC protocols for vehicular VLC (photodiode-based) to avoid problems such as hidden
terminal / expose terminal. Experiments will be carried out with off-the-shelf automotive lights
and our VLC receiver to determine appropriate decoding and interfering range. These values will
then be used in simulations to determine the improvement of throughput using different flavors
of MAC protocols.
2. More experiments will be carried out to evaluate the feasibility of the dual-camera CamCom
systems.