Biblio5

Figure 15.17 Trunk/ feeder system layout for two-way operation. (From Ref. 1, reprinted with permission.)

15.6.1.1 Comments on Figures 15.16a and 15.16b. Large guardbands isolate upstream from downstream TV and other services, 24 MHz in Figure 15.16a and 25 MHz in Figure 15.16b. A small guardband was placed in the frequency slot from 170 MHz to 174 MHz to isolate downstream voice and data signals from conventional CATV television signals.

We assume the voice service will be POTS (plain old telephone service) and that both the data and voice will be digital.

In another approach, downstream voice, data, and special video are assigned the band 550 MHz–750 MHz, which is the highest frequency segment portion of this system (Ref. 6). In this case, we are dealing with a 750-MHz system.

The optical fiber trunk terminates in a node or hub (see Figure 15.12). This is where the conversion from an optical signal to the standard CATV coaxial cable format occurs. Let a node serve four groupings of cable subscribers, each with a coaxial cable with the necessary wideband amplifiers, line extenders, and taps. Typically such subscriber groups would consist of 200 –500 terminations (TV sets) each. Assume each termination has upstream service using the band 5–30 MHz (see Figure 15.16a). In our example, the node has four incoming 5–30 MHz bands, one for each coaxial cable termination. It then converts each of these bands to a higher-frequency slot 25 MHz wide in a frequency division configuration for backhaul on a return fiber. In one scheme, at the headend, each 25-MHZ slot is demultiplexed and the data and voice traffic is segregated for switching and processing.

Access by data, voice, and special video users of the upstream and downstream assets is another question. There are many ways this can be accomplished. One unique method suggested by a consultant is to steal a page from the AMPS cellular radio specifications (see Chapter 16 for a discussion of AMPS). Because we have twice the bandwidth of an AMPS cellular operator, and because there are no handovers required, there are no shadowing effects and multipath (typical of the cellular environment); thus a much simpler system can be developed. For data communications, CDPD can be applied directly.7

7CDPD stands for cellular digital packet data.

464 COMMUNITY ANTENNA TELEVISION (CABLE TELEVISION)

Keep in mind that each system only serves 500 users as a maximum. Those 500 users are allocated 25 MHz of bandwidth (one-way).

An interesting exercise is to divide 25 MHz by 500. This tells us we can allot each user 50 kHz full-period. By taking advantage of the statistics of calling (usage), we could achieve a bandwidth multiplier of from 4× to 10× by using forms of concentration. However, upstream video, depending on the type of compression, might consume a large portion of this spare bandwidth.

There are many other ways a subscriber can gain access. DAMA techniques, where AMPS cellular is one, are favored. Suppose we were to turn to a digital format using standard 8-bit PCM. Allowing 1 bit/ Hz and dividing 25 MHz by 64 kHz, we find only some 390 channels available. Keep in mind that these simple calculations are not accurate if we dig a little further. For instance, how will we distinguish one channel from another unless we somehow keep them in the frequency domain, where each channel is assigned a 64-kHz slot? This could be done by using QPSK modulation, which will leave some spare bandwidth for filter roll-off and guardbands.

One well-thought-out approach is set forth by the IEEE 802.14 (Ref. 7) committee. This is covered in Section 15.7.

Why not bring fiber directly into the home or to the desk in the office? The most convincing argument is economic. A CATV system interfaces with a home/ office TV set by means of the set-top box. As discussed earlier, the basic function of this box is to convert incoming CATV channels to a common channel on the TV set, usually either channel 2, 3, or 4. Now we will ask much more of this “box.” It is to terminate the fiber in the AM system as well as to carry out channel conversion. The cost of a set-top in 1999 dollars should not exceed $300. With AM fiber to the home, the cost target will be exceeded.

The reason that the driving factor is the set-top box is the multiplier effect. In this case we would be working with multipliers of, say, 500 (subscriber) by >$300. For a total CATV network, we could be working with 100,000 or more customers. Given the two-way and digital options, both highly desirable, the set-top box might exceed $1000 (1997 dollars), even with mass production. This amount is excessive.

15.6.2 Impairments Peculiar to Upstream Service

15.6.2.1 More Thermal Noise Upstream Than Downstream. Figure 15.17 shows a hypothetical layout of amplifiers in a CATV distribution system for two-way operation. In the downstream direction, broadband amplifiers point outward, down trunks, and out distribution cables. In the upstream direction, the broadbnd amplifiers point inward toward the headend and all their thermal noise accumulates and concentrates at the headend. This can account for from 3 dB to 20-dB additional noise upstream at the headend, where the upstream demodulation of voice and data signals takes place. Fortunately, the signal-to-noise ratio requirements for good performance of data and voice are much less stringent than for video, which compensates, to a certain extent, for this additional noise.

15.6.2.2 Ingress Noise. This noise source is peculiar to a CATV system. It basically derives from the residence/ office TV sets that terminate the system. Parts 15.31 and 15.35 of the FCC Rules and Regulations govern such unintentional radiators. These rules have not been rigidly enforced.

One problem is that the 75-Q impedance match between the coaxial cable and the TV set is poor. Thus not only all radiating devices in the TV set, but other radiating

devices nearby in residences and office buildings couple back through the TV set into the CATV system in the upstream direction. This type of noise is predominant in the lower frequencies, that band from 5 MHz to 30 MHz that carries the upstream signals. As frequency increases, ingress noise intensity decreases. Fiber-optic links in an HFC configuration provide some isolation.

15.7 TWO-WAY VOICE AND DATA OVER CATV SYSTEMS ACCORDING TO THE IEEE 802.14 COMMITTEE STANDARD

15.7.1 General

The narrative in this section is based on a draft edition of IEEE Standard 802.14 dated March 11, 1997 (Ref. 7) and subsequent narrative kindly provided by the chairman of the IEEE 802.14 committee (Ref. 8). The model for the standard is a hybrid fiber-coaxial cable (HFC) system with a service area of 500 households. The actual household number may vary from several hundred to a few thousand, depending on penetration rate.

Two issues limit the depth and completeness of our discussion: (1) reasonable emphasis and inclusion of details versus page count, and (2) because of the draft nature of the reference document, many parameters have not been quantified. Figure 15.18 is a pictorial overview of the 802.14 system.

The IEEE 802.14 specification supports voice, data, video, and file transfer and interactive data services across an international set of networks. These are represented by switched data services such as ATM (asynchronous transfer mode; Chapter 18), variable length data services such as CSMA/ CD (Section 11.6.2.1), near constant bit rate services such as MPEG digital video systems (Chapter 14), and very low latency data services such as virtual circuits or STM (synchronous transfer mode; (e.g., E1 and DS1 families covered in Chapter 6). Instantaneous data rates and actual throughput are no longer limited by the protocol, but are, rather, a function of the network traffic engineering and theoretical limit of the media and the modulation schemes that are employed. This

Figure 15.18 A pictorial overview of the IEEE 802.14 networks and their relationship with the outside world. (From IEEE 802.14, Draft R2, p. 14 [Ref. 7].)

466 COMMUNITY ANTENNA TELEVISION (CABLE TELEVISION)

generates a wide range of QoS (quality of service) parameters that must be supported simultaneously in order to create a scaleable, multiservice delivery system.

The 802.14 system can be envisioned as having OSI layers 1 and 2. Layer 1, of course, is the physical layer, and layer 2, from a LAN viewpoint, covers the “lower portion” of the data link layer. This portion includes the functions of medium access control (MAC), which is briefly discussed in Section 15.7.2. The “upper portion” of the data link layer is the LLC (logical link control), described in Section 11.5.3. In the following discussion, we will use the acronym PHY for the physical layer.

15.7.2 Overview of the Medium Access Control (MAC)

As overall controller of the actual transmission and reception of information, the MAC must account for the unique physical topology constraints of the network while guaranteeing the required QoS for each type of data to be transported. The network consists of a multicast or broadcast downstream from the headend to individual subscriber groups and multiple allocated and contention upstream channels. The downstream channel consists of a single wideband, high symbol rate channel, composed of six-octet time allocation units. A single unit can be assigned an idle pattern or multiple units can be used to create ATM cells, variable length fragments, or MPEG video streams.

The 802.14 system allows for multiple simultaneous downstream channels as well. The upstream consists of multiple channels divided in time into a series of minislots. These represent the smallest orthogonal unit of data allocation. There is enough time in one minislot for the transmission of eight octets of data plus PHY overhead and guard time. Multiple minislots can be concatenated in an upstream channel to create larger packet data units such as ATM cells, variable length fragments, or even MPEG video streams. Because of the varying amount of overhead and guard time required by different physical layers, the number of minislots required for the transmission of any data stream will vary from one upstream PHY to another.

The use of minislots with independent in-channel/ in-band control messaging creates a flexible architecture. This flexibility allows one to change traffic flow patterns of the network and to fully integrate multiple channels and time slots.

15.7.3 Overview of the Physical Layer

Similar to the MAC architecture, the physical topology of the hybrid fiber-coax (HFC) plant allows for multicast downstream and multiple converging upstream paths. The 802.14 specification does not specify data types and resident topology. Constraints are defined and resolved while allowing the architecture to adapt on a session-by-session basis. Two distinctly different downstream PHYs are supported. Each type is centered around an existing coding and modulation standard: ITU-T Rec. J.83 (Ref. 9) Annex A/ C, which is adopted for European cable systems, and Annex B, which is adopted for North American cable systems. In addition to these standards, 802.14 specifies modulation, coding sequence, scrambling method, symbol rates, synchronization, physical layer timing, message length and formats, transmitter power, and resolution characteristics.

15.7.3.1 Subsplit/ Extended Subsplit Frequency Plan. The majority of CATV systems, particularly in North America, are upgraded to subsplit8 (5 MHz to 30 MHz) or

8Subsplit is a frequency division scheme that allows bidirectional traffic on a single cable Reverse path signals come to the headend from 5 MHz to 30 MHz, and up to 42 MHz on newer systems. Forward path (downstream) signals go from the headend to end-users from 54 MHz to the upper frequency limit of the system in question.

468 COMMUNITY ANTENNA TELEVISION (CABLE TELEVISION)

Figure 15.20 Downstream message hierarchy. I indicates and idle pattern. (From IEEE Std. 802.14, Draft R2, Figure M-2, [Ref. 7].)

single network may have several upstream channels to which stations may be assigned. Each station must be capable of changing its upstream channel at the request of the HC. The HC must be able to simultaneously receive all upstream channels within the network.

15.7.5.2 MAC Framing and Synchronization. All stations must be slaved to the master timing source that resides at the HC. To provide a time base to all stations that is synchronized correctly, the HC broadcasts a time-stamped cell to all stations at periodic intervals (2 ms). The HC can adjust each station’s timebase through messages so that all stations are synchronized in time. For any two stations on the network, it is important that, if they both decide to transmit at a given network time, both transmissions will arrive at the HC at the same instant.

15.7.5.3 Channel Hierarchies

15.7.5.3.1 Downstream Hierarchy. The downstream is composed of six-octet allocation units. A single unit can be assigned the idle pattern or multiple units can be used to create ATM cells or variable length fragments. Some of the ATM cells will carry MAC messages in the form of information elements. All MAC messaging is done in ATM cells with certain header values described in the reference publication. The downstream hierarchy is shown in Figure 15.20.

15.7.5.3.2 Upstream Channel Hierarchy. The upstream channel is a multiple access medium. For each upstream channel in the network there is a group of stations that share the assigned bandwidth. Each upstream channel is divided in time into a series of mini (time) slots. A minislot has the time capacity for the transmission of eight octets of data plus PHY overhead and guard time. A PDU (protocol data unit) that only occupies a single minislot is termed a minipdu. Minipdus are used primarily for contention opportunities to request bandwidth (bit rate capacity).

15.7.6 Physical Layer Description

15.7.6.1 Overview of the PHY. The PHY of 802.14 supports asymmetrical bidirectional transmission of signals in a CATV HFC network. The network is point-to-multi- point, tree branch access network in the downstream direction, and multipoint-to-point, bus access network in the upstream direction. The downstream transmission originates at the headend node and is transmitted to all end-users located at the tips of the branches in the tree and branch network. An upsteam transmission originates from an end-user

Figure 15.21 A model for HFC CATV serving area topology. (From IEEE Std. 802.14, Draft R2, Figure P-31 [Ref. 7].)

node and reaches the headend node through a multipoint-to-point access network where the access medium is shared by all end-users that are communicating with the same headend.

An example of a CATV HFC network topology is shown in Figure 15.21. In this case we see a multiple of 5–42 MHz upstream channels that are frequency division multiplexed in the fiber node (FN) and that are then transmitted via a single fiber trunk to the CATV headend. This operation is called frequency stacking.

470 COMMUNITY ANTENNA TELEVISION (CABLE TELEVISION)

15.7.6.2 Downstream Physical Layer Specification. There are two distinctly different PHYs supported by the 802.14 standard. These PHYs are called type A and type B downstream PHYs. The principal difference between the two is the coding method used for FEC (forward error correction; see Section 10.5.3.2). The FEC for type A downstream is based only on RS (Reed-Solomon) block coding. On the other hand, the FEC for type B downstream PHY is based on a concatenated coding method with outer RS coding and inner trellis-coded modulation (TCM). Each downstream PHY type has different modes of operation.

15.7.6.2.1 Downstream Spectrum Allocation. The frequencies from 63 MHz up to the upper frequency limit supported by the CATV cable plant (e.g., 750 MHz) are allocated for downstream transmission. Within this band is a channelized approach (i.e., frequency slots 6 MHz or 8 MHz wide) from the headend to end-user nodes. Standard CATV frequency plans are assumed. The topology model of the system is shown in Figure 15.21.

15.7.6.2.2 Propagation Delay and Delay Variation. The propagation delay for optical fiber is nominally 5 ms/ km and for coaxial, 4 ms/ km. The propagation delay introduced by the downstream transmission medium should be budgeted such that the total round-trip delay between the heandend and the end-user station should be a maximum TBD (to be determined) milliseconds.9

15.7.6.3 Type A Downstream PHY. The Type A downstream PHY supports two modes of operation: 64-QAM and 256-QAM. A block coding approach based on the shortened RS coding is used. A convolutional interleaver mitigates the effects of burst noise.

15.7.6.3.1 Constellations for Type A Downstream PHY. Figure 15.22 illustrates a type A PHY constellation for 64-QAM and 256-QAM waveforms.

15.7.6.4 Type B Downstream PHY. Type B downstream PHY supports two modes of operation: 64-QAM and 256-QAM. As previously mentioned, the coding strategy is different for type B downstream.

15.7.6.5 Downstream Carrier Frequencies. The downstream carrier frequencies are selected in accordance with the following:

f c c (n × 250 kHz) ± 8 kHz,

where n is an integer such that 63 MHz ≤ f c ≤ 803 MHz.

15.7.6.6 Transmitted Signal Levels. The type B downstream PHY is capable of transmitting a signal on the coaxial cable between the ranges of +50 dBmV and +61 dBmV.10

9If the round-trip (loopback) system extension is 160 km (quoted from above) and seven-eights of the system is optical fiber, then the optical fiber portion is 140 × 5 ms + 40 × 4 ms, for a total of 0.760 ms.

10We can get away with calling a voltage a level because in this case the impedance is always assumed to be 75 Q .

Figure 15.22 Constellations for type A downstream PHY. (a) 64-QAM for type A downstream PHY; (b) 256-QAM for type A downstream PHY. (From IEEE Std. 802.14, Draft R2, Figure P-37 [Ref. 7].)