In the telecom boom, small-to-medium-sized businesses (SMBs) got left in the lurch. With so many requests in hand, operators dedicated most of their attention to meeting the demands of large corporations and end users, forcing SMBs to fall behind on the broadband revolution.
But, with the market declining, operators have rethought their position on SMBs, making smaller corporations a main target for new bandwidth services. To attract these customers, however, carriers must effectively build broadband solutions that deliver fast, secure pipes to these customers.
To up broadband performance in SMB apps, operators and equipment should consider a network topology that marries the hybrid fibre coax (HFC) network through fixed broadband wireless technology. This will examine a proposed HFC/broadband wireless approach, looking at the benefits and challenges this technology delivers. To kick off the discussion, let's start with an overview of the HFC architecture.
Typical HFC Architecture
Figure 1 shows a simplified diagram of a typical HFC network employing a tree-and-branch-type topology. The downstream broadcast signal, consisting of analog and digital video, is sub-carrier multiplexed to form a composite signal typically between 55 to 870 MHz. The composite signal then amplitude modulates a particular wavelength, usually in the 1550-nm band.
Click here for Figure 1
Simplified block diagram of an HFC network.
The narrowcast signals in an HFC network, meant for specific geographical regions, mostly contain of DOCSIS (cable modem), video-on-demand (VOD), and similar signals. These are sub-carrier multiplexed quadrature amplitude modulated (QAM) signals, typically between 450 to 870 MHz. In the HFC architecture, these narrowcast signals amplitude modulate multiple wavelengths in the 1550-nm band for transport purposes.
In the HFC structure, several narrowcast wavelengths are multiplexed together with broadcast signals using DWDM technology for transport over the fiber network. The multiplexed signal is transmitted over the fiber to a hub, where all but one of the narrowcast wavelengths is dropped. The remaining narrowcast wavelength is then multiplexed back together with the broadcast signal and re-launched to one of the destination fiber optic nodes.
The nodes convert the optical signals to electric signals, and distribute them to the homes using the coax plant. Similarly, the return signal (typically 5 to 45 MHz, mostly DOCSIS) is converted to light at the node and eventually reaches the Headend. There may be several stages of processing to the return signal before it is converted to light.
When looking at a HFC architecture, it's important to note that the fiber portion is under-utilized since it carries much less effective data compared to a telecommunication network carrying the signals in the native baseband format. As an example, the downstream data traffic (excluding analog video) in a HFC plant seldom reaches 1.5 Gbit/s per wavelength. For telecommunications networks, speeds of 2.5 Gbit/s are routine and new deployments tend to go for 10 Gbit/s per wavelength.
The number of wavelengths is also a great deal less in the HFC network. The number of wavelengths rarely reaches 16 in a HFC fiber plant. Telecom networks, on the other hand, often deliver 40 or more wavelengths, each carrying 2.5-Gbit/s payloads.
Optimizing for Data
Until now, operators have relied on a traditional HFC networks, which are optimized for video delivery, for providing data services. Over the past few years, however, a host of companies have pitched additional HFC solutions that are more data optimized. These include passive optical networks (PONs), proprietary (non-DOCSIS) QAM over coax, and proprietary time division multiplexing (TDM) of Fast Ethernet signals.
The problem with most of these new solutions, however, is the reliance on wiring to the SMBs. For example, PONs require operators to rollout fiber links all the way to the customer premises. While the price of the fiber cable is on the decline, extending the reach of the fiber to the customer premises can be very costly and time consumingnot good things in today's cost-sensitive communication sector.
To counteract the problems associated with delivering broadband services over wires, some operators have pitched fixed broadband wireless technology, such as LMDS and MMDS, as a means for bridging the last-mile datacom gap. Under these topologies, operators tried to set up a single point-to-multipoint (PMP) microwave radio system located at the Headend that would serve a 20-km radius (MMDS). Some tried using mm-wave radios (LMDS) from the Node onwards. In either case QAM signals (DOCSIS-based) already present in the fiber were made to radiate.
But while LMDS and MMDS eliminate the wiring challenges, they bring their own set of headaches. In particular, many of these systems rely on line-of-sight operation, which is difficult to achieve. Additionally, these systems are severely impacted by multipath interference problems. The DOCSIS PHY also caused some challenges in this design. This PHY was not designed for free-space transmission. Therefore, we needed additional equalization schemes when building these radios. The inherent limitations of DOCSIS such as timesharing of upstream bandwidth were present, though the radio physical layer (PHY) did not create a bandwidth bottleneck. Since cable operators do not have a license for this spectrum, this can cause additional headaches.
What's needed is a compromise. HFC has done a nice job extending the reach of the fiber plant closer to the customer premise. Wireless provides advantages on the wiring front. If the two can be combined together, then designers can more easily bridge the last mile gap. Specifically, an architecture is being proposed that will place small, low-cost radios at the Nodes of the HFC network and thus deliver data to SMBs and other end users. Below we'll detail the hybrid fiber wireless (HFW) architecture.
The proposed HFW architecture uses the fiber part of the HFC network and bypasses the coax bottleneck with a last mile wireless network. Since it is not necessary to use any RF carriers when the last mile is not coax, the most efficient way to run data signals in the fiber is in the native baseband format. Thus, there is an overlay of baseband signals over legacy analog signals in the fiber part of the network. As explained below, this can be achieved by the proper use of DWDM technique. In this manner, the utilization of the fiber plant is greatly enhanced, and an order of magnitude increase in throughput can be achieved. Thus, bypassing the last mile coax not only removes a bandwidth bottleneck, but also allows baseband transmission on a lighted fiber carrying legacy signals.
Figure 2 shows a set of business getting connected to the HFC network using a broadband wireless link. In this figure, signals are being sent over an RF link to a data node. Once these signals are received, they are transmitted over the HFC fiber plant to the headend using wavelengths different from those used to carry the legacy traffic, like narrowcast signals. To multiplex these signals together, designers need to employ a hybrid DWDM (HDWDM) technique, which combined analog and digital signals together over a signal fiber link. This DWDM technique is protocol-agnostic, thus allowing it to handle a host of traffic schemes such as Gigabit Ethernet, Sonet, Fibre Channel, and more.
Click here for Figure 2
Figure 2: Wireless bypass of coax and wavelengths carrying baseband data.
Figure 3 provides a deeper look at the HFW topology, using Gigabit Ethernet as an example. In this figure, legacy equipment is colored green and the additional equipment in brown. We note that the additional equipment saves considerable expenditure by bypassing the cable modem termination system (CMTS) and linear transmitters, replacing them with low-cost baseband transmitters and receivers.
Click here for Figure 3
Figure 3: Typical headend design in an HFW system.
Figure 4 shows how the narrowcast (QAM) and baseband signals may be interleaved. The launch power of the baseband wavelengths can be lower than the narrowcast wavelengths by several dB, and therefore, the existing optical amplifiers may not need replacement.
Click here for Figure 4
Figure 4: Interleaving of narrowcast and baseband signals.
Figure 5a shows the wavelength add-drop scheme at the hub in downstream direction. As before, the legacy equipment is colored green and the additional equipment in brown. In the upstream, the concentration is achieved by straightforward DWDM multiplexing of the baseband wavelengths from different nodes. This multiplexed optical signal can be combined (or interleaved) with the legacy wavelengths. It is assumed that each data node is assigned a separate baseband wavelength. The additional optical add-drop multiplexer (OADM) in Figure 5a needs to be as low insertion loss as possible and is a challenge. The same applies to the OADM in the data node (Figure 6).
Click here for Figure 5a
Figure 5a: Drop and insertion of baseband wavelength at hub downstream.
Click here for Figure 5b
Figure 5b: Drop, multiplexing and insertion of baseband wavelengths at hub upstream.
Figure 5b shows the DWDM multiplexing at the hub for upstream traffic. One (or more) wavelength(s) is assigned to each data node for upstream traffic. The return path usually carries one legacy wavelength (analog or digitized analog), while the baseband from the data node can combined optically with an optical tap as shown in Figure 6. The baseband is dropped at the hub, multiplexed with wavelengths from other data nodes, and inserted back to the fiber reaching the headend.
Click here for Figure 6
Figure 6: Layer 2 switch in HFW data node.
Figure 6 shows the dropping and insertion of the baseband wavelength from the data node and the Layer 2 Ethernet switch inside the data node. The 10/100 ports can be interfaced to the radio as shown in Figure 7. The CPE side is typically interfaced to a fat client-server that might serve a LAN within an enterprise or a multi-dwelling unit (MDU), or might serve multiple customers with individual billing mechanisms.
Click here for Figure 7
Figure 7: PP radio interfaced to data node and CPE.
Since this solution is intended for campus and SMB applications, multiple point-to-point (PP) radios, rather than a single PMP, are used. PP radios are employed because SMBs and MDUs occur in clusters as opposed to residential neighborhoods, which tend to be uniformly distributed over an area. In principle, any PP radio operating in any band (including free-space optics) and following any modulation scheme can be used. From practical considerations, however, the motivation is to use modulation schemes like OFDM to successfully combat the LOS and multipath issues. Additionally, designers must employ radios that operate in unlicensed bands since cable operators do not have access to licensed bands, like LMDS.
The use of PP radios provides several benefits in the HFW architecture. First, these radios are less susceptible to in-band interference from other users. Additionally, these radios deliver higher throughput. Total throughput from all the PP links served by a data node will exceed the data throughput from a PMP radio. Power amplification of OFDM signals is also handled more easily due to the fact that antenna gain helps in achieving a requisite EIRP. Also a PP radio provides scalability in terms of CAPEX and disruption of a single link does not affect the services in others
Benefits of HFW
The HFW architecture provides many benefits to the design. First, it eliminates the bandwidth and asymmetric data rate problems encountered in traditional HFC networks. The use of PP radios between the headend and CPE provides guaranteed bandwidth. Additionally, HFW uses the fiber bandwidth optimally. To achieve comparable throughputs with conventional means (QAM-modulated carriers and linear transmitters), much more power needs to be launched into the fiber. As the number of wavelengths keep going up, there may be serious fiber non-linearity issues.
The HFW architecture also drops cost in the broadband. Through this approach, the headend can interface directly to a router, thus eliminating the costly CMTS.
The HFW also does not jeopardize legacy services. Additional fibers are not needed and the newly added equipment coexists with the current equipment.
Some Challenges to Overcome
There are some issues that designers must deal with when developing the HFW architecture. Clearly, the goal of HFW is to provide additional data services through the HFC plant without causing degradation in the legacy services. The most sensitive signals carried through the HFC network are the downstream analog signals, where stringent requirements of received optical power and carrier-to-noise ratio (CNR) need to be satisfied. The additional OADMs in the HFW topology can introduce as much as 1-dB of insertion loss, hurting the transmission of downstream analog signals. Designers can compensate for this problem by increasing the power output of the downstream optical amplifier
The baseband signal between the headend and data node also suffers more optical loss than the legacy signals in the HFW approach. A loss budget exercise using typical receiver sensitivity numbers demonstrates the power requirement of the optical transmitters to be within reasonable limits.
Crosstalk between wavelengths carrying baseband and QAM signals can also cause headaches for developers. Wavelength separation need to be contained within acceptable limits so as not to degrade signal-to-noise ratios (SNRs) of adjacent channels. Typical QAM wavelengths are spaced 1.6 nm apart. Thus baseband wavelengths interleaved between QAM ones with state-of-the-art DWDM devices
should not introduce unacceptable crosstalk in HFW networks.
Future Development Activities
Since the HFW architecture is relatively new, work must be done to improve efficiency in the topology. For example, it is likely that the same customer premises might be covered by more than one data node. This opens up the possibility of redundant path switching to increase the reliability of the network. Some form of antenna beam steering may be necessary in such a situation.
A by-product of the HFW could be the location of the "Edge QAM" device in the data node. This device would convert the baseband data into QAM format for DOCSIS protocol and utilize the coaxial plant. This would eliminate the linear transmitters in the headend carrying QAM and increase network throughput.
Replacement of the multiple PP radios with an omni-directional base station would clear the way towards roaming capability. Since the omni stations would be connected to the headend through fiber, various capabilities, like handoff and billing, could be handled from a centralized location.
Author's Note: The author would like to acknowledge the contributions of Ajay Das of Onirban Networks, Ashok Kumar of Sohoware, Dr. Amit Sen of UBS AG, and Dr. Deven Verma of TiE.
- A. Sen, A Method of Providing Very High Bandwidth Information Transport over Hybrid Fiber Coax (HFC) Networks, Document Disclosure to US Patent Office, April 03, 2000.
- Pan, Jin-Yi, Hybrid Analog/Digital WDM Access Network With Mini-Digital Optical Node, United States Patent 6,147,786, November 14, 2000.
- S. Mukherjee and A. Sen, High Speed Data Services using Hybrid Fiber Coax Network, US Patent Application 10116390, April 02, 2002.
About the Author
Somnath Mukherjee has over 15 years of extensive engineering background and is the founder of RB Technology, a consulting company. Somnath was a key technical leader in the development of systems and subsystems for HFC networks at Silicon Valley Communications Inc. He has also worked in the areas of radio mesh networks, high-speed modems for microwave radios, wireless closed-circuit TV systems, and precision RF/microwave test instruments. He has an MSEE from Syracuse University and a Doctor of Engineering from the University of Kansas. Somnath can be reached at firstname.lastname@example.org.