WO1988001066A1

WO1988001066A1 - Optical fiber distribution network

Info

Publication number: WO1988001066A1
Application number: PCT/US1986/001628
Authority: WO
Inventors: Bruce D. Campbell; Joseph Zucker
Original assignee: Raychem Corporation
Priority date: 1986-08-06
Filing date: 1986-08-06
Publication date: 1988-02-11
Also published as: FI872658A; HUT47187A; FI872658A0

Abstract

An optical fiber distribution network (2) includes non-destructive read taps (23) disposed in series which create extremely small attenuations of an optical signal (31) being tapped, and which increase the bandwidth of an optical fiber (12) being tapped. The network further includes non-destructive write taps (22) disposed in another series which each inject an optical signal (131) into a second optical fiber (11) while inducing minimal attenuation of a pre-existing optical signal (37) in the second fiber (11).

Description

OPTICAL FIBER DISTRIBUTION NETWORK The present invention relates to novel optical read and write taps for reading and writing information into and out of an optical fiber, methods enabled thereby, and networks rendered useable therewith.

As used throughout herein, "optical fiber" refers to a waveguide having a core surrounded by a cladding, preferably surrounded also by a buffer, the core and cladding being glass, either multimode or single mode, preferably the core, cladding, and buffer each being cylindrical.

Numerous methods have been proposed in the prior art for creating optical fiber distribution networks, and a basic disadvantage of all prior art solutions is that taps for injecting light into and withdrawing light from an optical fibers used in such networks disadvantageously create large attenuations to signals within the optical fiber such that only relatively few subscribers can be connected to a network per optical repeater spacing. For example, read taps generally induce 1 dB attenuation, with write taps generally inducing an attenuation of 2 dB or more. Since numerous optical repeaters are required to make such networks operational, the cost of the network is rendered unacceptably high making all prior art proposals of academic interest only.

It is a first essential object of the invention to provide a method and means in series for withdrawing a small part 25 (see Figure 1) of an optical signal 31 out of an optical fiber, the part 25 being large enough so that data represented by this signal can be detected or read, the part 25 being withdrawn so that an attenuation of the signal 31 in the fiber is minimized. A second essential object is to provide a method and means in series for multiplexing optical signals into an optical fiber in a way such that minimal attenuation to any optical signals already in the fiber occurs.

A third essential object is to provide the method and means used to achieve the first and second essential objects set forth above so as to form an optical fiber distribution network capable of both reading and writing, so that data, audio, and video information, for example, can be transmitted.

A fourth essential object is to provide a method and means for withdrawing a small part 25 of an optical signal 31 out of an optical fiber in a manner such that a bandwidth of the part 25 of the signal withdrawn is greater than a bandwidth of the signal 31 if the signal 31 were detected at an end of the fiber at a power level equal to a magnitude of the signal part 25.

These and other objects will be further described with reference to the accompanying figures of which:

Figure 1 is a schematic view of a light read tap of the invention;

Figures 2 and 3 illustrate cumulative and non-cumulative excess losses, respectively, created by sequential read taps attached to a fiber in series;

Figures 4 and 5 illustrate cumulative and non-cumulative losses, respectively, of further read taps disposed in series on an optical fiber; Figure 6 illustrates a comparison between the cumulative loss shown in Figure 4 as a function of cumulative bend sector angle and the cumulative loss which results from a continuous fiber bend.

Figure 7 is a cross-sectional view of a fiber core showing an inner circular region thereof surrounded by an outer annulus section thereof;

Figure 8 is a graph illustrating an optical signal on- to-off ratio as a function of modulation frequency;

Figure 9 is a schematic view of a light write tap constructed according to the present invention;

Figures 10 and 11 illustrate cumulative and non- cumulative attenuations imposed on a preexisting optical signal by sequential light write taps disposed in series on an optical fiber; and

Figure 12 illustrates one preferred optical fiber distribution network architecture rendered useable by the read and write taps of the invention.

Referring to Figure 1, according to the invention, an optical fiber 12 having an optical signal 31 therein is tapped using a side read tap 23 so as to withdrawn a part 25 of the optical signal to be later detected. According to the invention, the tap 23 is constructed so that the excess loss, identified by reference number 31'' in Figure 1, is minimized so that an untapped optical signal 31' remaining in the fiber 12 downstream of the tap 23 is maximized.

As used throughout herein, the term "excess loss" means the magnitude of the signal 31 which is withdrawn from the optical fiber 12 and not detected, e.g. the difference between signal 31 and the sum of signal 31' and signal part 25 (31-31'-25).

Accordingly, assuming a magnitude of the optical signal part 25 is minimized but is large enough to enable detection, since the excess loss 31' is small, numerous taps 23 can be disposed in series so as to form an optical fiber read bus 3 or 5 (Figure 12), which has the advantage that numerous subscribers using the numerous taps 23 can share the cost of all the components forming the read bus. A preferred magnitude of the detected optical signal part 25 is between -20 dBm and -100 dBm, (+X dBm denoting a quantity equal to ±X dB above or below one millawatt), preferably between -30 dBm and -80 dBm, more preferably between -30 dBm and -40 -50, or -60 dBm, e.g. about -35 dBm.

Referring back to Figure 1, the tap 23 is formed so as to bend the fiber 12 in a region 46 about a bend region sector angle alpha and a uniform or non-uniform bend radius. For a non-uniform bend radius, as is shown in Figure 1, the minimum value of the radius is referred to by r, this letter denoting the radius of the bend region when it is uniform as well. An optical coupler 32 is disposed in contact with only a portion of an outer circumferencial bent surface of the fiber in the region 46. Optionally, a detector 34 and a lens 35 can be incorporated within the coupler 32. The detected signal is then relayed by any appropriate means, such as wires 44. In an alternative embodiment, the withdrawn optical signal part 25 is transmitted by another optical fiber having an end located in a vicinity of a front face 30 of the lens 35, this additional optical fiber extending to any desired destination, such as a remote optical component (e.g. a remote detector). In this embodiment, the additional optical fiber would contact the fiber buffer and would correspond to the optical coupler 32.

The optical coupler 32 preferably is transparent at wavelengths anticipated for the optical signal 31, can be a liquid or solid, or combinations thereof, and can be deformable or non-deformable, and when deformable can be either resilient or non-resiliently deformable, a preferred embodiment being a resiliently deformable material such as a transparent polysiloxane. If the coupler is another optical fiber, a front end thereof could contact the bus fiber 12 buffer. A cladding diameter of the additional fiber is preferably between 200 and 2000 um, preferably 500-2000 um, most preferably 1000 um, and the additional fiber can have a core and cladding made of either glass or plastic. It is desirable that the optical coupler make good surface contact with the fiber buffer, be optimally positioned, contact the fiber buffer only on a small outer circumferencial portion of the fiber buffer and over a small linear length 41 of the buffer, and have an index of refraction which optimizes a magnitude of the detected optical signal part 25, all other parameters being constant.

The bend region section angle alpha is the arc enclosing the bend region 46, which is determined by the intersection of first and second normal dash lines 47, 48 which are normal to and extend from points 43, 49 where the bent center axis, indicated by dash line 36, of the optical fiber 12 merges with first and second unbent center axes 28, 29 of the optical fiber 12 at opposite sides of the bent region 46.

The relatively sharp bend in the vicinity of the point 49 has been exaggerated for clarity of illustration. When the bend region 46 is non-uniform in profile, optimum results are obtainable if a minimum bend radius r of the bend region 46 occurs at a location 24 disposed upstream of the lens 35 or detector 34, or alternatively the additional optical fiber. All types of bend radius profiles across the bend region 46, including both uniform and varying profiles, are included within the scope of the invention.

In general, if the optical signal part 25 is orders of magnitude less than the optical signal 31, then the attenuation of the signal 31 is little effected by the withdrawn part 25, with the number of taps being limited to essentially the magnitude of the excess loss 31'' created by each tap. Accordingly, it has been found that if alpha, -θ-, r, the bend radius profile, and the size, shape, composition, index of refraction, and location of the optical coupler are optimized, the magnitude of the withdrawn optical signal part 25 can be controlled to be any desired small value, and it has also been surprisingly found that the excess loss 31'' which results thereby is also surprisingly and unexpectedly small such that numerous taps 23 can be disposed in series, with each tap being capable of determining information represented by the optical signal 31.

Example 1

A graded index optical fiber having a core diameter of 100um, a cladding diameter of 140um, and a buffer diameter of 500um was spliced so as to create a 2.2 km link, and to one end of the fiber a laser was attached with the result that 8.42 mw was detected at an opposite second fiber end by a photodetector. 219 read taps as illustrated in Figure 1, having a graded index lens 35 and a photodetector 34, were attached to the optical fiber at 10 m spacings sequentially from the laser and adjusted so as to each detect 1 uw. The optical power at the opposite. second end of the optical fiber was sequentially noted, and the total and excess loss created by each read tap was determined.

The first few taps (e.g. first 5 or 10) had extremely low excess losses, averaging less than 0.05 dB, the excess losses of taps downstream therefrom being even less, the average total loss for the 219 tap experiment being 0.0217 dB.

As can readily be appreciated from Figures 2 and 3, the side light read taps of the invention can withdraw and detect a detectable optical signal part 25 so as to create an excess loss of well under 0.3 dB per tap, specifically less than 0.2 dB, more specifically less than 0.1 dB, and values even less than 0.08 dB, 0.06 dB, 0.04 dB, 0.03 dB, 0.02 dB, 0.01 dB and 0.005 dB. Even with such low excess losses, always the optical signal part 25 withdrawn for detection has a detectable magnitude, such as any value between -20 and -100 dBm as indicated previously, the value used during the experiment being -30 dBm. Accordingly, large numbers of taps can be disposed in series, such as 5, 10, 20, 30, 40, 50, 70, 100, 150, 200, and even numbers in excess thereof, and any number therebetween, using conventional light sources since an ample amount of light is available for the last tap of the series to withdraw a detectable optical signal part 25 so as to occur a minimal excess loss.

This is surprising since it is known that the amount of light capable of being withdrawn at a fiber bend of constant radius at one location is limited, this phenomena surprisingly not being applicable to sequential bends created by taps of the invention separated in space, for example separated by more than one meter, five meters, ten meters, 20 meters, 30 meters, 40 meters, 70 meters, or 100 meters. This is illustrated in Figures 4-6.

Specifically, Figures 4 and 5 illustrate the cumulative and non-cumulative losses, respectively, created by 100 constant radius 30° bends of radius 3.84 mm sequentially applied to the fiber of Example 1 at distances separated by 10 meters. The "30° taps" data points in Figure 6 illustrate this same cumulative loss as a function of cumulative bend sector angle. The "continuous wrap" data points in Figure 6 illustrate the cumulative loss created when the fiber of Example 1 was simply continuously wrapped a plurality of times about a radius equal to 3.84 mm. As Figure 6 illustrates, when a fiber is continuously wrapped more than a few turns, additional bends of constant radius create no appreciable additional light loss, thus establishing an arrangement of the taps 23 of the invention in a long series to be totally unexpected.

When the data on the read taps of the invention was further analyzed, it was discovered that the performance of the read taps of the invention actually improves when disposed in a series arrangement downstream from other taps in the manner described in Example 1 and shown in Figures 2-5, e.g. the excess loss of the read taps sequentially becomes lower while detected optical signal parts are always withdrawn, this surprising effect having been discovered to be the result of the upstream taps modifying an energy distribution of the optical fiber within the fiber in a beneficial way.

Specifically, referring to Figure 7, if a fiber core is divided into a center region 94 having a diameter 95 equal to the core radius, and a uniform annulus 96 having a uniform width 97, 98 equal to half the fiber radius, it is possible to determine the percentage of total light energy in the core center 94 and the core annulus 96. Upstream of the taps in Example 1, 60% of the energy was contained wi thin the core center 94 with the remaining 40% of the energy being in the core annulus 96. After six taps in series, these percentages changed to 81% and 19% respectively. Though it would normally be expected these percentages would soon change to 100% and 0% respectively (and hence allow no further signal parts 25 to be withdrawn), since progressive taps continue to withdraw light from the core annulus 96 preferentially to the core center 94, these percentages stayed at 81% and 19% respectively even after 106 taps. Accordingly, it is seen that the taps of the invention modify an energy distribution in the fiber in a beneficial way, so as to both reduce the excess loss for taps located downstream, and to regenerate light in the core annulus 96 as well to allow the downstream taps to always be capable of withdrawing detectable optical signal parts 25 while creating a progressively lower excess loss.

Though these experiments were conducted on multimode graded index fiber, similar beneficial results are achieved as well for step index and single mode fiber, the invention in its various embodiments being applicable to all kinds of fiber with various kinds of buffers, e.g. silicone or acrylate, since taps of the invention can continuously withdraw detected optical signal parts 25 in series while creating very low excess losses on all types of fiber.

A further advantage of the read taps of the invention when used on multimode fiber, either step index or graded index, is that any one of the read taps of the invention operating alone or in conjunction with other similar taps actually improve the bandwidth of the fiber since only outermost modes are preferentially withdrawn as opposed to innermost modes. Accordingly, time of arrival dispersion between various modes is reduced, this increasing the bandwidth.

The detrimental effect of time of arrival dispersion between various modes in a multimode optical fiber on bandwidth is well understood in the art, and accordingly will only be briefly described. Specifically, since outermost modes must travel a larger linear distance in a fiber than innermost modes, a generated square pulse is seen at a remote end of the fiber as a dispersed pulse since the innermost modes arrive at a detector at the remote end prior to the outermost modes, and hence the pulses must be widely separated in time resulting in a low bandwidth. Since the taps of the invention do not detect appreciable amounts of innermost modes, the time of arrival dispersion problem is reduced with the taps of the invention with the result that the square pulses can be grouped closer together thus resulting in a higher bandwidth.

More specifically, referring to Figure 8, any given signal intensity level results in a constant signal on-to- off ratio for a range of modulation frequencies until the modulation frequency reaches a saturation modulation frequency 55, and thereafter the signal on-to-off ratio falls rapidly as the modulation frequency is further increased. It has been found that, for any given initial signal on-to- off ratio, the saturation modulation frequency is much higher when using taps of the invention than when employing end detection techniques. In fact, it has been found that at relatively high modulation frequencies, if a detector detected end light in the fiber at a power level light is detected by the tap of the invention so as to experience a second signal on-to-off ratio, with the tap of the invention experiencing a first signal on-to-off ratio, the second signal on-to-off ratio is more than 1 dB, 2 dB, 3 dB, 4 dB, 5 dB, and even 10 dB lower than the first signal on-to-off ratio.

A further feature of the invention is shown in Figure 9. This figure illustrates a light write tap 22, a plurality of which are illustrated in Figure 12, this write tap being of construction similar to the read tap 23 previously described. The write tap 22 is designed to be capable of injecting an optical signal 131 into a fiber so as to be multiplexed with a preexisting optical signal 37 previously injected into the fiber, the preexisting signal either comprising a single signal or a pluraly of multiplexed signals. As is well known, multiplexed signals can be frequency multiplexed and/or time multiplexed, all such variations being included within the term "multiplexed signals".

The write tap 22 is designed to inject the new signal 131 so as to minimize an attenuation of the preexisting signal 37, which has the advantage that numerous write taps 22 can be disposed in series on the fiber 11 with the preexisting signal 37 not being unduly attenuated so that numerous subscribers using the taps can share the cost of the distribution network formed thereby. The part of the preexisting signal 37 which is not attenuated by the write tap 22 is illustrated by arrow 37', the attenuated part of the preexisting signal being represented by arrow 37''.

The write tap 22 includes means for bending the fiber 11 so as to create a fiber bent region 146, and means for contacting an optical coupler 132 against an outermost layer of the fiber buffer at a coupling area 141 on an outside portion of the bent region 146. The optical signal 131 to be injected into the optical fiber 11 is generated by light source 134, preferably a laser or a light emitting diode (LED), and is preferably focused by a lens 135, preferably a graded index lens, so as to propagate within the optical coupler 132 at an angle such that it will enter the fiber 11 within an angle of acceptance thereof and be bound within the fiber core, the bound signal being illustrated by arrow 131 at the right side of Figure 40.

The purpose and function of the optical coupler 132 is similar to that of the coupler 32, that is, to optimize reflective and refractive effects at the coupler-buffer interface and to optimize light injection into the fiber core. Without the coupler 132, substantial amounts of the signal 131 would reflect off the outermost surface of the buffer or be refracted into the fiber 11 at angles outside a cone of acceptance of the fiber core created by the bent region 146. The optical coupler 132 can have a material composition and construction identical to the coupler 32, or can be different therefrom.

As with the coupler 32, the material forming the.optical coupler 132 should have optimum transparency characteristics, be capable of making good surface contact with the buffer at the buffer coupling area 141, have an index of refraction which optimizes the light injection, and be optimally positioned such that θ is optimal, all materials described as being suitable for the coupler 32 being usable for the coupler 132 and being included within the scope of the invention. A bend region section angle is defined as an arc enclosing the bent region 146, which is determined by the intersection of first and second normal dash lines 147, 148 which are normal to and extend from points 143, 149 where the bent center axis, indicated by dash line 136, of the optical fiber 11 merges with first and second unbent center axes 128, 129 of the optical fiber 11 at opposite sides of the bent region 146.

The relatively sharp bend in the vicinity of the point 149 has been exaggerated for clarity of illustration. The bent region 146 does not necessarily have to be bent about a uniform radius throughout, and in fact the radius of curvature of the bent region 146 can be non-uniform. When the bend radius is non-uniform, optimum results are obtainable if a minimum bend radius R of the bent region 146 occurs at a location 124 disposed downstream of the lens 135 or light source 134. All types of bend radius profiles across the bent region 146, including both uniform and varying profiles, are included within the scope of the invention.

The invention includes a method and means of injecting the optical signal 131 into the optical fiber so as to create an unexpected and surprisingly small attenuation 37" of the upstream preexisting signal 37, the preexisting signal having been previously injected into the optical fiber using side or end launch techniques. Attenuations less than 1.0 dB, 0.5 dB, 0.4 dB, 0.3 dB, 0.2 dB, 0.1 dB, and 0.06 dB, and even less than 0.04 dB have repetitively been accomplished in accordance with the invention, while achieving injection efficiences greater than 1%, 2%, 5%, 10%, 15%, 20%, 30%, and even 40%. Hence, a ratio of a percentage of light injected to a percentage of the preexisting signal attenuated can be made greater than 1, preferably greater than 1.5, 2, 2.5, 3, 4, 5, 6, 10, 15, 20, 30 and even 40.

It has long been known that a fundamental problem with multiplexed optical fiber bus architectures carrying multiple optical signals generated at multiple positions and injected into the fiber at multiple points is that the means used for injecting any one of a plurality of optical signals into an optical fiber already carrying a pre-existing optical signal therein necessarily attenuates some percentage of the pre-existing optical signal, and that the percentage of attenuation increases with the injection efficiency of the injecting means. A common rule of thumb in the art is that normally there is a one to one relationship between the injection efficiency and the percentage of attenuation of the pre-existing signal already in the fiber, e.g. a 90% injection efficiency from a source generating a first signal causes 90% of the pre-existing signal to be attenuated, a 50% injection efficiency causes 50% attenuation, a 10% injection efficiency causes 10% attenuation, etc. There are minor deviations from this rule of thumb for the particular injection means chosen, but these deviations have not been very significant to date.

It has unexpectedly and surprisingly been discovered that a side light write tap of the invention is capable of producing an injection efficiency which is far higher than a percentage of attenuation induced to a pre-existing signal in the fiber, if the tap 22 is used in a specific manner.

Specifically, if N additional means 22, which are similar in construction to a write tap 22 shown in Figure 40, are disposed between an upstream position where the preexisting optical signal is injected into the fiber and a downstream position where an Nth + 1 write tap injects its Nth + 1 signal into the fiber, with N being an integer, even for relatively low values of N, the injection efficiency/ attenuation ratio for the Nth + 1 tap 22 can be dramatically increased, this ratio becoming extremely large for values of N greater than about 10-15, with significant improvements occurring for values of N as small as 1, 2 and 3. In addition, though the N intervening means 22 produce some loss, the overall average loss per means 22 quickly falls to relatively low values far below the average otherwise obtained by application of other prior art method and means for light injection in series.

The means 22 can comprise a series of additional write taps 22, can comprise a series of bending elements which bend the fiber in a manner similar to the write tap 22 so as to cause mode coupling within the core, or can comprise corny binations of such write taps and bending elements. If bending elements are used, additional side write taps which inject light into a side of the fiber at a straight section can be used if disposed in close proximity upstream of a bending element since the bending element will downward couple cladding and/or buffer modes into the fiber core. The use of additional write taps allows for signal multiplexing. Also, the Nth + 1 tap can be a tap which injects light into a fiber at a side bend, as illustrated in Figure 40, or can inject light at a side at a straight fiber section with the fiber bend being immediately downstream thereof to achieve mode coupling.

An unexpected and surprising result is that a ratio between the injection efficiency of the write tap 22 to an attenuation percentage of the signal 37 increases as the number N increases . Example 2

A write tap of Figure 9 was attached to a graded index optical fiber of the type used in Example 1, this tap injecting about 30% to 40% of its power into a core of the fiber to create a preexisting signal therein. Subsequently, 112 additional write taps were sequentially attached at ten meter intervals from the first tap downstream thereof to the fiber so as to be each capable of injecting about 30% to 40% of their outputs into the fiber core, and the cumulative and non-cumulative loss to the preexisting signal was sequentially noted, the results being set forth in Figures 10 and 11, respectively.

As these figures indicate, the first downstream write tap attenuated a preexisting signal 2.75 dB (46:9%), the second 1.91 dB (35.6%), the third 1.37 dB (27.1%), etc., the progressive attenuation to the preexisting optical signal rapidly falling such that after the twelfth downstream write tap attenuations were consistently under 0.2 dB even though injection efficiencies remained between 30% and 40%, attenuations less than 0.06 dB and 0.04 dB being repetitively achieved. In fact, the average attenuation for all the taps was 0.2 dB, despite the rather large excess losses incurred by the first several taps.

From this data it has been determined that upstream taps attenuate the preexisting optical signal more than twice the amount of downstream taps for identical injection efficiencies for the first few downstream taps, this ratio rising to four times , six times, ten times, fifteen times, twenty time, thirty times, and more, as further taps are disposed downstream, this being a dramatically surprising and unexpected non-linear phenomena. Again, it was determined the first several taps 22 modify an energy distribution of the preexisting optical signal in the optical fiber in a beneficial way to create these results such that it has been determined 5, 10, 30, 50, 100, and even 200 write taps, or any number therebetween, can be disposed on a fiber for multiplexing a similar number of signals into a fiber and yet one is still capable of detecting the preexisting signal passing through all these taps. Though the preexisting signal has been described as being in jected into a core of the fiber through a bent buffer section thereof, similar advantageous results have been obtained with end launch signals as well. In each case, it should be noted the write taps inject signals into the fiber advantageously by passing the light through the fiber buffer at a bend in a non-destructive way such that the fiber does not need to be terminated, the fiber sections being maintained substantially continuously bent by a radius sufficiently large such that the fiber does not break.

It has further been determined that optical fiber minimum bend radii used with read or write taps of the invention create surprising and unexpected low probabilities of fiber breakage such that it is possible to utilize read and write taps of the invention on an optical fiber so as to substantially continuously permanently maintain various sections of the optical fiber in a bent attitude with minimum risk that any one of the fiber sections will fail by fracturing caused by strain. In fact, it has theoretically and experimentally been determined that a permanent bend in an optical fiber of the type of the previous examples which has a minimum bend radius equal to 3.5 mm or larger over a 45° sector angle creates less than a 10% chance of fracturing the fiber section in 20 years, and a 3.8 mm or larger minimum bend radius over 45° creates less than a 1.5x10^-2% chance of fracturing the fiber section in 20 years. Accordingly, as used herein, the term "substantially continuously maintained in a bent attitude" means that the fiber is maintained bent continuously over a substantial period of time, in particular in excess of a month, preferably in excess of 1 year, and even in excess of 5, 10, 15, and 20 years, preferably such that a probability of failure of the bent section is less than .1, more preferably less than 1.5x10"2. Minimum bend radii equal to or in excess of^" 4.2 mm and 4.5 mm result in much lower probabilities of fiber breakage over extended periods of time, and are suitable for use with read and write taps of the invention.

Figure 12 illustrates a preferred distribution architecture for the read and write taps 23, 22 of the invention. In this figure, subscribers 6, 6' and 10 are interconnected in a network 2 by multiplexingly injecting signals into a write bus 4 by using write taps 22, as described, and reading optical signals from one or more read buses 3, 5 using read taps 23, as described. In this case, separate read buses may, but not necessarily want to be provided for audio and video information» Lines (either optical or electrical) 20, 18 transmit signals out of and into the network 2, with a CPU 17 and optical light sources 14, 16 initially creating optical signals which are injected into the read buses 3, 5 which can then be tapped.

The CPU also appropriately allocates time slots or frequencies for the various subscribers 6, 6¹ , 10 so that each subscriber has some dedicated frequency or time slot within which to read and write information so that information from a pluralilty of subscribers is not superimposed and thereby destroyed. The network 2 is capable of serving hundreds of subscribers without the need of any significant number of optical repeaters due to the minute amounts of excess losses and attenuations imposed by the read and write taps 23, 22, as described. It can be appreciated that the bus architecture of Figure 12 can be disposed in a line, branch, tree, star, or any other desired topology.

Though the invention has been described by reference to certain preferred embodiments thereof, the invention is not to be limited to only those described embodiments, and is to be limited only by the appended claims.

Claims

1. A method of reading information represented by an optical signal in an optical fiber at a bend, the fiber including a core, a cladding, and a buffer, comprising the steps of:

withdrawing a first part of an optical signal through a side of an optical fiber and through its buffer at a fiber bend so as to create an optical signal excess loss less than 0.3 dB; and

detecting the first part of the optical signal.

2. The method of claim 1, the excess loss being less than 0.2 dB, preferably less than 0.1 dB, more preferably less than 0.06 dB, and most preferably less than 0.02 dB.

3. The method of claim 1 or 2, further comprising the step of withdrawing N additional parts of the optical signal through a side of a fiber and its buffer at N additional fiber bends so as to create an average optical signal excess loss less than 0.3 dB, preferably less than 0.2 dB, more preferably less than 0.1 dB, most preferably less than

0.06 dB, N being an integer greater than 1, N preferably being greater than 10, the first and N additional parts of the optical signal being withdrawn from the optical fiber in a series arrangement; and

detecting the N additional parts of the optical signal.

4. The method of claim 1 or 2, further comprising the step of modifying an energy distribution of the optical signal within the optical fiber upstream of a first location where the first part of the optical signal is withdrawn so as to reduce an integrated energy ratio between an outermost and an innermost section of a core of the optical fiber at the first location, the energy distribution being modified by bending and withdrawing N additional parts of the optical signal and detecting the N additional withdrawn optical fiber parts, N being an integer greater than 1, N preferably being greater than 10.

5. The method of claim 3 or 4, N being an integer greater than 10, preferably greater than 20, more preferably greater than 30, 40, 50, 100, or 200.

6. An apparatus constructed to carry out the method of any one of claims 1-5.

7. The apparatus of claim 6, the apparatus withdrawing all of the signal parts through the optical fiber buffer without requiring termination of the optical fiber and without damaging the fiber or its buffer.

8. A method for multiplexing optical signals into an optical fiber so as to reduce an attenuation of a pre-existing optical signal already in the optical fiber, comprising the steps of:

disposing N light write taps in series at N optical fiber sections downstream of a point where the preexisting optical signal is injected into an optical fiber, the N write taps injecting N optical signals into the fiber, N being an integer, an Nth tap being disposed upstream of other ones of the N taps, the first tap being disposed downstream of other ones of the N taps, a ratio of an injection efficiency percentage of a first injected optical signal created by the first tap to an attenuation percentage of the pre-existing optical signal caused by the first tap being greater than 1, preferably greater than 1.5.

9. The method of claim 8, the ratio being greater than 2, preferably greater than 2.5, 3, 4, 5, or 6, more preferably greater than 10, even more preferably greater than 15, 20 or 30.

10. A method for multiplexing a first optical signal at a first location into an optical fiber using a first light write tap so as to reduce an attenuation of a pre-existing optical signal in the optical fiber caused by the first light write tap, comprising the steps of:

modifying an energy distribution of a pre-existing optical signal within an optical fiber upstream of a first location so as to reduce an integrated energy ratio between an outermost and an innermost section of a core of the optical fiber in a vicinity of the first location; and

injecting the first optical signal into a first optical fiber section through a side thereof at the first location, an attenuation of the pre-existing optical signal created by the first light write tap being less than an attenuation of the pre-existing optical signal if the integrated energy ratio was not modified.

11. The method of claim 10, the first light write tap injecting the first optical signal into the first optical fiber section through its buffer, the energy ratio being modified by multiplexingly injecting N additional optical signals into N additional optical fiber sections at N additional locations through a side of the N additional optical fiber sections using N write taps, the N additional optical signals being injected into their respective optical fiber sections through optical fiber buffers thereof upstream of the first tap.

12. The method of any one of claims 8, 9, and 11, N being greater than 5, preferably greater than 10, more preferably greater than 30, and most preferably greater than 50.

13. The method of any one of claims 8, 9, 11, and 12, the attenuation of the preexisting optical signal caused by the first tap being less than 0.3 dB, preferably less than

0.1 dB, more preferably less than 0.06 dB or 0.04 dB.

14. The method of any one of claims 8, 9, 11, 12, and 13, all of the N write taps being non-destructive and injecting the N optical signals into the N fiber sections through buffers thereat without requiring termination of the optical fiber and without requiring the fiber or its buffer being damaged.

15. An apparatus utilizing the method of any one of claims 8-14.

16. An optical fiber distribution network, comprising:

N means for injecting and multiplexing N optical signals into N optical fiber sections at N first locations in a first optical fiber link such that the N optical signals pass through a bent side of the N sections, respectively, the N sections being substantially continuously maintained in a bent attitude sufficient for continuously injecting the N signals into the first optical fiber link, N being a first integer; M means for withdrawing M parts of an additional optical signal out of M optical fiber sections at M second locations in a second optical fiber link such that the optical signal parts pass through a bent side of the M sections, respectively, the M sections being substantially continuously maintained in a bent attitude sufficient for continuously withdrawing the M optical signal parts, M being a second integer; and

means for interconnecting the first and second optical fiber links.

17. The network of claim 16, the interconnecting means including means for allocating time slots to the N injection means and the M withdrawing means.

18. A method of increasing bandwidth of an optical fiber capable of transmitting information by supporting light pulses in a plurality of modes, the fiber including a core, a cladding, and a buffer, comprising the steps of:

injecting light pulses into the optical fiber;

withdrawing a portion of the light pulses through the fiber buffer at a first location with a non-destructive light read tap;

detecting light withdrawn by the light read tap using a detector thereby obtaining a detected first power level and a first signal on-to-off ratio;

the light pulses being injected into the optical fiber at a modulation frequency in excess of a saturation modulation frequency, the saturation modulation fre quency being that frequency whereby the detector would incur a second signal on-to-off ratio which is less than the first signal on-to-off ratio if the light pulses were detected by the detector at an end of the fiber at the first location at the first power level.

19. The method of claim 18, the second signal on-to-off ratio being more than 1 dB lower than the first signal on- to-off ratio, preferably more than 2, 3, 4, 5 or 10 dB lower.

20. The method of claim 18 or 19, the bit error rate resulting from the first signal on-to-off ratio being less than the bit error rate resulting from the second signal on- to-off ratio.