STI: Multi-Hop Free Space Optics Last-Mile Networks Using Very Low-Cost Components

Rensselaer Polytechnic Institute, Troy NY

Investigators

Abstract

The tremendous explosion of bandwidth in the core of the Internet has stopped short of end-users due to the well-known "last-mile" problem where it has traditionally been infeasible to get a high bandwidth conduit to the home and businesses. "Fiberless" or Free Space Optical (FSO) networks can effectively complement unlicensed-spectrum RF-based WLAN technologies like 802.1 lb/a, and WMAN technologies like 802.16 for the "last mile" problem. Free space optics offers tremendous bandwidth and potential for dense spatial reuse of the optical spectrum. Although optical networking has become popular in the wired networking world, free space optics (FSO) is still a niche technology used to provide only selected point-to-point links, and not as a general-purpose metropolitan area networking technology. Moreover, several fundamental aspects of free space terrestrial optical communication are yet to be explored. The proposal is to investigate fundamental limits of very low-cost components, and create communication systems that tap into the high bandwidth potential of FSO, while addressing its key limitations (eg: Line-of-Sight (LOS) availability, LOS discovery, LOS alignment). An investigation of very low-cost communication systems and multi-hop optical networks using high-brightness visible spectrum LEDs (HBLEDs) that have more attractive propagation & spatial reuse characteristics compared to ordinary brightness infrared LEDs will occur. Moreover the potential size, weight, power and form-factor aspects of LED-based systems would make it ideal for use in ad-hoc infrastructures (eg: balloons, treetops, lampposts. poles on chimneys, moving vehicles etc), and novel applications like high-bandwidth mobile ad-hoc networks and future sensor networks may be found. Using trans-receiver pairs that achieve 100 Mbps with 1-10 mW power, the proposal is to use dense integration of thousands of trans-receivers into 2-dimensional and 3-dimensional spatial structures (eg: 2-d array, 3-d spherical/honeycomb structure). The 2-d array would provide extremely high aggregate bandwidth (100 Gbps and more) over 1-2 km. The proposed 3-d spherical structures (a.k.a. optical antenna) combined with LOS auto-discovery and LOS auto-tracking techniques to instantly discover and track LOS if it exists. This structure will allow uninterrupte4,jiltra-high-speed operation even in high velocity sway (eg: 120 mph, range over 100s of meters) or even mobile conditions. The project will build low-cost FSO interconnection systems, complemented with relatively cheap and programmable electronic components after l0s-l00s of opto-electronic (OEO) hops. It is also proposed to investigate spatial coding and redundancy techniques at the electronic hops taking advantage of the particular spatial structures (eg: 2d array, spherical etc). An examination of fundamental network capacity limits for FSO-based fixed and ad-hoc configurations, and fundamental power-per-bit achievable in future sensor networks will be done. Evaluating the impact of such multi-hop FSO networks on the performance of TCP/IP, VoIP and multimedia applications. Leveraging a partnership with renowned Internet entrepreneur, field trials of a mix of our prototype low-cost FSO-based systems, off-the-shelf 802.1 lx systems, IP routing and 802. ld bridging protocols will occur. The proposed inter-disciplinary research will make significant contributions to LED-based communications, low-cost free space optical networks, and models for deploying last-mile networks in unlicensed spectra. NSF recognizes the need to solve the lost mile problem and funds this project with the hope of a solution immediately emerging from this project, or new directions to a solution being established.