`
`
`
`1111111111111011111111111111011!1121111°111,111)!) ,1,111111111011110111111111111111
`
`(19) United States
`(12) Patent Application Publication (10) Pub. No.: US 2002/0122413 Al
`Sep. 5, 2002
`Shoemake
`(43) Pub. Date:
`
`(54)
`
`ADAPTIVE FRAGMENTATION FOR
`WIRELESS NETWORK COMMUNICATIONS
`
`(52) U.S. Cl.
`
`370/349; 370/470
`
`(75)
`
`Inventor: Matthew B. Shoemake, Allen, TX
`(US)
`
`(57)
`
`ABSTRACT
`
`Correspondence Address:
`TEXAS INSTRUMENTS INCORPORATED
`P 0 BOX 655474, M/S 3999
`DALLAS, TX 75265
`
`(73) Assignee: Texas Instruments Incorporated, Dal-
`las, TX
`
`(21) Appl. No.:
`
`10/026,088
`
`(22) Filed:
`
`Dec. 21, 2001
`
`Related U.S. Application Data
`
`(60) Provisional application No. 60/262,507, filed on Jan.
`18, 2001.
`
`Publication Classification
`
`(51) Int. Cl.'
`
`H04J 3/24
`
`A wireless local area network (LAN) adapter (20) that
`optimizes the length of message packets, for example
`according to the IEEE 802.11 standard, and in an environ-
`ment having interfering transmissions (BL1 et seq.), is
`disclosed. The disclosed adapter (20) executes an adaptive
`process by way of which an adjustment to the packet length
`is derived based upon rate measures for the most recent two
`trial packet lengths. The rate measure corresponds to a
`packet success rate for that packet length, determined either
`from an estimating function or by actual measurements,
`multiplied by a ratio of the data portion of each packet to a
`total packet length including interpacket spacing. Upon
`convergence as the adjustment becomes smaller, the opti-
`mized packet length for best data rate given the present
`interference. A method of determining the need for packet
`length optimization is also disclosed, in which the actual
`packet error rate is compared against an expected packet
`error rate based upon signal-to-noise ratios.
`
`40
`
`INITIALIZE µ
`
`/-44
`
`SELECT TIMES //
`t0, t h
`
`INITIALIZE PACKET
`LENGTH t0
`
`/ J- 56
`
`MEASURE PACKET
`SUCCESS RATE FOR /
`CURRENT tpk
`
`INCR k
`
`DERIVE RATE MEASURE /
`FE FOR CURRENT t k
`P,
`
`45
`
`,,— 4B
`
`5C
`
`52
`
`DERIVE A = FE -
`
`/— 54
`
`SELECT NEXT
`
`tpk + µA
`
`55
`
`YES
`
`-.<7,NT1NUE?
`
`NO
`
`.
`
`END
`
`Sonos Ex. 1010, p. 1
` Sonos v. Google
` IPR2021-00964
`
`

`

`Patent Application Publication
`
`Sep. 5, 2002 Sheet 1 of 6
`
`US 2002/0122413 Al
`
`/-
`
`VVL2
`
`10
`
`12
`
`INTERNET
`
`BL4
`
`- 14
`
`WL1
`
`BL1
`
`5
`
`Fla
`
`4
`
`- BL2
`
`7
`
`BL3
`
`01E
`
`6
`
`Sonos Ex. 1010, p. 2
` Sonos v. Google
` IPR2021-00964
`
`

`

`Patent Application Publication
`
`Sep. 5, 2002 Sheet 2 of 6
`
`US 2002/0122413 Al
`
`A
`
`25
`
`26
`
`PHY
`
`20
`
`22
`
`PCB
`
`RADIO
`
`COMPUTER
`
`N
`
`I/F
`
`MAC
`
`23 -N
`
`27
`
`24
`
`CPU
`
`MEMORY
`
`FIG, 2
`
`Sonos Ex. 1010, p. 3
` Sonos v. Google
` IPR2021-00964
`
`

`

`Patent Application Publication
`
`Sep. 5, 2002 Sheet 3 of 6
`
`US 2002/0122413 Al
`
`/
`
`PR
`
`H
`
`PL
`
`PREAMBLE
`
`HEADER
`
`PAYLOAD
`
`C
`R
`C
`
`tH
`
`to
`
`to
`
`tp = tH +to
`
`FIG. 3
`
`Sonos Ex. 1010, p. 4
` Sonos v. Google
` IPR2021-00964
`
`

`

`Patent Application Publication
`
`Sep. 5, 2002 Sheet 4 of 6
`
`US 2002/0122413 Al
`
`F 1G. 4A
`
`q'
`
`tp (µsec)
`
`R (Mbps)
`
`FIG. 4B
`
`R'
`
`tp,k+1
`
`tp (µsec)
`
`Sonos Ex. 1010, p. 5
` Sonos v. Google
` IPR2021-00964
`
`

`

`Patent Application Publication
`
`Sep. 5, 2002 Sheet 5 of 6
`
`US 2002/0122413 Al
`
`30
`
`32
`
`34
`
`FIG. 5
`
`RECEIVE SNR, SINR
`FROM RECEIVED
`PACKETS
`
`NO
`
`DETERMINE EXPECTED
`PACKET ERROR RATE pe
`
`V
`DERIVE ACTUAL PACKET
`ERROR RATE p FROM CRC
`CHECKS
`
`36
`
`40
`
`P-Pe > 89
`
`YES
`
`V
`
`OPTIMIZE PACKET
`LENGTH
`
`Sonos Ex. 1010, p. 6
` Sonos v. Google
` IPR2021-00964
`
`

`

`Patent Application Publication
`
`Sep. 5, 2002 Sheet 6 of 6
`
`US 2002/0122413 Al
`
`40
`
`INITIALIZE
`
`44
`
`SELECT TIMES /
`to, th
`
`V
`INITIALIZE PACKET
`LENGTH t
`
`P
`
`MEASURE PACKET
`SUCCESS RATE q' FOR
`CURRENT tpk
`
`56
`
`INCR k
`
`DERIVE RATE MEASURE
`F k FOR CURRENT tpk
`
`46
`
`48
`
`50
`
`DERIVE A = F k - Fk l
`
`r -- 52
`
`Fla 6
`
`54
`
`SELECT NEXT tpk+1 = t pk + µA
`
`YES
`
`/-
`
`55
`
`CONTINUE?
`
`NO
`
`IP- END
`
`Sonos Ex. 1010, p. 7
` Sonos v. Google
` IPR2021-00964
`
`

`

`US 2002/0122413 Al
`
`Sep. 5, 2002
`
`1
`
`ADAPTIVE FRAGMENTATION FOR WIRELESS
`NETWORK COMMUNICATIONS
`
`CROSS-REFERENCE TO RELATED
`APPLICATIONS
`
`[0001] This application claims priority, under 35 U.S.C.
`§119(e), of provisional application No. 60/262,507, filed
`Jan. 18, 2001.
`
`STATEMENT REGARDING FEDERALLY
`SPONSORED RESEARCH OR DEVELOPMENT
`
`[0002] Not applicable.
`
`BACKGROUND OF THE INVENTION
`
`[0003] This invention is in the field of wireless commu-
`nications, and is more specifically directed to the wireless
`transmitting of data packets in an environment containing
`possible interfering communications.
`
`[0004] Wireless local area networks (LANs) have become
`increasingly popular in recent years. Typically, wireless
`LAN installations include an access point sited within the
`vicinity of the various client workstations. The access point,
`which is typically a network element coupled to a computer
`workstation by way of Ethernet cabling or the like, serves as
`a hub for wireless devices within its communications range.
`Bidirectional communications are carried out between the
`access point and wireless network-enabled devices that are
`in range (typically on the order of 100 m), enabling the
`wireless devices to communicate with one another, with
`other computers resident on the same wired network as the
`access point, and with remote computers over the Internet.
`
`[0005] Under current wireless networking technology and
`standards, an example of such being the IEEE 802.11b
`standard, the wireless communications are packet-based, in
`that each transmission is transmitted in the form of multiple
`packets. By being packet-based, the packets need not be
`transmitted or received in sequence, and will generally not
`be contiguous in time. Indeed, as known in the art, packets
`that are corrupted in transmission are retransmitted later in
`time. Upon receipt of all of the packets for a communication,
`the receiver resequences and combines the packets into a
`coherent message. In the 802.11 context, each message
`packet typically includes a preamble and header portion that
`contains control information and also information identify-
`ing the packet (identifying the message, the sequence of the
`packet in the message, source and destination nodes, etc.),
`and also includes a payload portion that contains the actual
`data being communicated, along with a checksum by way of
`which errors in the payload portion can be detected and
`possibly corrected.
`
`[0006] Modern wireless networks typically operate in the
`unlicensed Industrial, Scientific, and Medical (ISM) band
`which, as known in the art, includes frequencies from about
`2400.0 MHz to about 2483.5 MHz. Conventional 802.11
`transmission s are signals according to the QPSK and BPSK
`constellations that are modulated into a "channel" within the
`ISM band having about a —20 dB bandwidth of about 16
`MHz, and providing data rates that can reach up to about 11
`MHz. Other wireless devices also communicate in this band.
`An example of such devices are the newly-popular "Blue-
`tooth" devices, which transmit in a frequency-hopping man-
`
`ner within the ISM band. More specifically, Bluetooth
`transmissions are carried out in channels that are about 1
`MHz in width (-20 dB) that change frequency periodically
`(e.g., about every 625 µsec).
`
`[0007] Considering the likelihood that both 802.11 and
`Bluetooth devices may be operating within the range of the
`802.11 access point, and also considering other ISM trans-
`missions such as wireless telephones, garage door openers,
`and the like, signal interference can often occur. If two
`different transmissions occur at the same time in the same
`frequency channels within the ISM band, typically both
`transmissions will be corrupted. Accordingly, those in the art
`have studied ways to reduce the incidence of collisions in
`this unlicensed band.
`
`[0008] Fragmentation is a conventional approach to reduc-
`ing the packet error rate due to interference. In general,
`fragmentation enforces an upper limit on packet length, thus
`reducing the likelihood that an interfering signal will occur
`within the packet. Typically, under the 802.11b standard, a
`parameter is used to set the number of payload data bytes
`transmitted in a packet for a given bit rate, which thus sets
`the packet length (or fragmentation level). As is fundamental
`in the art, because interference along any portion of the
`packet will corrupt the entire packet, longer packet lengths
`generally result in a higher probability of packet error due to
`interference, for a given level of interference. Increasing the
`fragmentation of the transmission, by using shorter payload
`portions in each packet, therefore provides a reduced packet
`error rate. However, because of the existence of a certain
`amount of overhead associated with the transmission of each
`packet the average throughput rate decreases as the packet
`lengths are decreased. Examples of such network overhead
`include packet preambles , packet headers and spacing
`between packets.
`
`[0009] This tradeoff between packet error rate and over-
`head makes the selection of a fragmentation in a wireless
`network an important constraint on the overall network
`performance. Making this selection even more difficult is the
`nature of the interferers that are now commonly present
`within a wireless network signal range. Many potential
`interferers, such as wireless telephones, garage door open-
`ers, and the like, may interfere only within certain times of
`operation. Other interferers, such as frequency-hopping
`transmissions in a Bluetooth network, further complicate the
`fragmentation selection, considering the ephemeral nature of
`the transmissions in the various channels. It can therefore be
`quite difficult to select a fragmentation level or a transmis-
`sion channel to maximize the average throughput rate.
`
`BRIEF SUMMARY OF THE INVENTION
`It is therefore an object of this invention to provide
`[0010]
`a method and a network element for wireless communica-
`tion of digital data in which the average throughput rate is
`maximized.
`
`It is a further object of this invention to provide
`[0011]
`such a method and element in which the optimizing of the
`average throughput rate can adapt to the environment in the
`frequency band being used.
`
`It is a further object of this invention to provide
`[0012]
`such a method and element for detecting a condition in
`which packet error rates are predominantly due to interfer-
`ence.
`
`Sonos Ex. 1010, p. 8
` Sonos v. Google
` IPR2021-00964
`
`

`

`US 2002/0122413 Al
`
`Sep. 5. 2002
`
`2
`
`It is a further object of this invention to provide
`[0013]
`such a method in which the optimization can be effected at
`the transmitter, without requiring change in the characteris-
`tics of the receiver.
`
`[0014] Other objects and advantages of this invention will
`be apparent to those of ordinary skill in the art having
`reference to the following specification together with its
`drawings.
`
`[0015] This invention may be implemented by way of an
`adaptive optimization of the packet length for wireless
`transmissions, particularly in environments with interferers.
`According to the invention, a successful rate value is deter-
`mined for an iteration of a packet length. For example, the
`successful rate value may correspond to the product of the
`probability of communicating a packet without error, with a
`ratio corresponding to the payload length over the total
`transmission time for the packet. The next packet length
`iteration is then determined by applying a weighted differ-
`ence of the most recent successful rate values to the current
`packet length iteration.
`
`[0016] According to another aspect of the invention, a
`method of determining the effects of interference on packet
`error length involves the comparison of an expected packet
`error rate based on signal-to-noise characteristics of the
`channel with an actual packet error rate based on error
`detection at the receiver for the actual payload.
`
`BRIEF DESCRIPTION OF THE SEVERAL
`VIEWS OF THE DRAWING
`
`[0017] FIG. 1 is an electrical diagram, in block form, of
`a wireless local area network (LAN) in an environment with
`interfering transmissions such as from a group of Bluetooth
`devices.
`
`[0018] FIG. 2 is an electrical diagram, in block form, of
`a wireless network adapter within which the preferred
`embodiment of the invention may be implemented.
`
`[0019] FIG. 3 is a flow chart illustrating a method of
`determining the presence of packet error due to interference,
`according to the preferred embodiment of the invention.
`
`[0020] FIG. 4a is a qualitative plot of packet success rate
`versus packet length, under various conditions.
`
`[0021] FIG. 4b is a qualitative plot of data rate versus
`packet length for one of the conditions of FIG. 4a.
`
`[0022] FIG. 5 is a timing diagram illustrating the defini-
`tion of various times within the transmission of a packet,
`according to the preferred embodiment of the invention.
`
`[0023] FIG. 6 is a flow chart illustrating a method of
`optimizing packet length in an interfering environment,
`according to the preferred embodiment of the invention.
`
`DETAILED DESCRIPTION OF THE
`INVENTION
`
`this invention. Those skilled in the art having reference to
`this specification will readily comprehend, however, that
`this invention may also be used in connection with other
`packet-based communications applications, with particular
`benefit to those applications that have ephemeral and fre-
`quency-hopping interferers. Accordingly, it is contemplated
`that those skilled in the art will recognize that the following
`description is presented by way of example only.
`
`[0025] FIG. 1 illustrates an example of a wireless LAN
`environment into which the preferred embodiment of the
`invention is implemented. As is fundamental to those in the
`art, wireless LAN environments can vary widely from
`installation to installation, and indeed can vary widely over
`time within a single installation as different devices arc
`installed, used, or enter and exit the signal range of the
`wireless LAN. Accordingly, the environment of FIG. 1 is
`presented by way of example only.
`
`In FIG. 1, computer 2 is enabled to carry out
`[0026]
`wireless LAN communications to and from wireless LAN
`access point 10 by way of wireless IAN adapter 20, which
`transmits and receives signals according to the IEEE 802.11
`standard over wireless link WL1. FIG. 2 illustrates an
`example of the construction of wireless LAN adapter 20, in
`cooperation with computer 2.
`
`In the example of FIG. 2, wireless LAN adapter 20
`[0027]
`includes host interface 22, which controls communications
`with computer 2 over bus PCI. Host interface 22 commu-
`nicates with medium access controller (MAC) 25, which is
`a conventional controller known in the art. Embedded CPU
`23 and off-chip memory 24 cooperate with MAC 25, to
`effect control of adapter 20 and to provide additional pro-
`gram memory, respectively. Physical layer (PHY) device 26,
`also referred to as a baseband processor, performs conven-
`tional modulation of data signals from MAC 25 into a spread
`spectrum form for transmission via radio circuitry 27, and
`demodulation of spread spectrum signals received from
`radio circuitry 27 into baseband. Antenna A is coupled to
`radio circuitry 27, in the conventional manner. In this
`example, interface 22, MAC 25, PHY 26, CPU 23, and some
`or all of memory 24 may be implemented into a single
`integrated circuit, such as the ACX100 spread spectrum
`processor with medium access control, available from Texas
`Instruments Incorporated. Of course, other architectures
`may alternatively be used for wireless LAN adapter 20. The
`arrangement of FIG. 2 is presented merely by way of
`example.
`
`[0028] Computer 2 is also connected to its various input
`and output devices by way of different facilities. For
`example, computer 2 is hardwire connected to its monitor 4
`by a conventional VGA cable. Computer 2 is also in wireless
`communication with printer 5, keyboard 6, and mouse 7, by
`wireless links BL1, BL2, BL3, respectively; in this example,
`links BL1 through BL3 effect communications according to
`the Bluetooth standard, which is a well-known short range
`frequency-hopping communications standard.
`
`[0024] Referring now to the Figures, an exemplary imple-
`mentation of this invention in connection with a wireless
`local area network (LAN) will now be described. As will
`become apparent, this invention is particularly beneficial
`when applied
`to wireless networks, considering
`the
`increased average data throughput rate that is achievable
`over a wide variety of time-varying conditions, by way of
`
`[0029] The wireless I,AN of the example of FIG. 1 also
`includes other workstations and network services. In this
`example, workstation 8 is also connected to access point 10
`by 802.11 wireless link WL2 via a wireless LAN adapter
`(not shown), while server 12 is hardwire connected to access
`point 10 by an Ethernet or other facility. Workstation 8 also
`shares printer 5, by way of Bluetooth wireless link BL4. In
`
`Sonos Ex. 1010, p. 9
` Sonos v. Google
` IPR2021-00964
`
`

`

`US 2002/0122413 Al
`
`Sep. 5. 2002
`
`3
`
`this example, server 12 is connected to modem 14, by way
`of which Internet access is provided to server 12, and also
`to computer 2 and workstation 8 over wireless links WL1,
`WL2, respectively.
`
`[0030] As evident from the example of FIG. 1, the mul-
`tiple wireless communications provide significant opportu-
`nities for interference. The 802.11 wireless links WL1, WL2
`must, of course, be coordinated so as to not interfere with
`one another; typically, access point 10 controls the channels
`(in frequency, or alternatively in time-multiplexed fashion)
`assigned to these links to avoid interference. However, non
`802.11 wireless activity, such as Bluetooth wireless links
`BL1 through BL 3, cannot be controlled in frequency or time
`by access point 10. Rather, according to this preferred
`embodiment of the invention, the presence of these inter-
`ferers is detected and, in that event, the packet lengths of the
`wireless LAN links WL1, WL2 are optimized for maximum
`data rate.
`
`[0031] As discussed above in the Background of the
`Invention, the likelihood that a packet will be interfered with
`increases with increasing packet length; conversely, if one
`assumes no interference, the transmission data rate will
`decrease with decreasing packet length because of packet
`overhead. According to this invention, a peak data rate as a
`function of packet length can be derived for a given inter-
`ference probability, as will now be described.
`
`[0032] FIG. 3 illustrates the construction of a message
`packet, such as may be transmitted and received according
`to the IEEE 802.11 standard. The packet includes preamble
`PR, which includes control and other information regarding
`the channel, specifically including a current measurement of
`the signal-to-noise ratio (SNR) and signal-to-interfering-
`noise ratio (SINR) under the 802.11 standard. Each packet
`also includes header II, which contains information identi-
`fying the message with which the packet is associated, the
`sequence of this packet in that message, the source and
`destination addresses of the packet, the packet length, and
`the like. Payload portion PL of course contains the data
`being transmitted in the packet, and includes CRC checksum
`by way of which the receiver detects, and possibly corrects,
`one or more bit errors in payload portion PL.
`
`[0033] By way of definition, the duration of preamble PR
`and header H portions of the packet occupies time tH, and is
`considered to be fixed among all packets. Payload portion
`PL, including its CRC checksum, has a duration ti,), and will
`be varied, according to the preferred embodiments of the
`invention, for optimum effective data rate. As known in the
`art, the 802.11 standard (as well as other standards), enforce
`a minimum interpacket spacing to, which will also be treated
`as a fixed quantity in this description. By way of definition,
`packet length tp will refer to the sum of times tE, and to. It
`will be understood by those skilled in the art having refer-
`ence to this specification that the various times within the
`packet may be defined differently, and used in optimizing the
`packet length, while not departing from the spirit of this
`invention.
`
`[0034] FIG. 4a illustrates various possible shapes of a
`success function q, which is the probability that a packet is
`transmitted without error in a given environment that
`includes interfering transmissions, as it varies with packet
`length tp. As shown in FIG. 4a, a minimum packet length tp
`exists, corresponding to a packet having one byte of payload
`
`portion PL in combination with the preamble PR and header
`portions II; this minimum of course will be briefly (one byte)
`longer than time tH. As discussed above, the success rate is
`maximum for the shortest packets, considering that the
`probability of interference with a packet is lower as the
`packet size decreases. The particular relationship of success
`function q with increasing packet length tp may take various
`forms, as suggested by FIG. 4a, including linear and higher
`order forms. While it is contemplated that a linear approxi-
`mation q' will usually resemble this relationship, the present
`invention is equally applicable to other orders of success
`function q.
`
`[0035] For a given success function q(t0), one can derive
`the effective data rate R as a function of packet length tp:
`
`R -
`
`
`
`Xr
`t + to
`
`x q(tp) -
`
`(rj, —ryixr
`
`+ to
`
`x q(tp)
`
`[0036] where r is the bit rate of data transmission in
`payload portion PL of the packet. For the case of linear
`success function q' of FIG. 4a, the data rate function R' as
`a function of packet length tp is plotted in FIG. 4b. As
`evident from FIG. 4b, data rate function R' has a single
`maximum. According to the preferred embodiment of the
`invention, an adaptive approach is performed to determine
`this maximum in operation, and to set the packet length tp for
`each transmission to achieve this maximum.
`
`It is possible that no interferers are in the vicinity
`[0037]
`of wireless LAN adapter 20 and its fellow LAN elements.
`Knowledge of whether or not an inteferer is in the vicinity
`may be useful for various purposing such as to notify a user,
`to provide input to interference avoidance algorithms, to
`assist networks administrators and to determine whether or
`not adaptive fragmentation is useful due to the presence of
`interference. According to the preferred embodiment of the
`invention, therefore, a method for determining whether
`interference is present would be useful. This detection
`method will now be described with reference to FIG. 5.
`
`[0038] The method of FIG. 5 and the other operations of
`LAN adapter 20 described in this specification are prefer-
`ably executed by programmable logic within LAN adapter
`20, for example by embedded CPU 23. Of course, if MAC
`25 or baseband processor 26 has sufficient processing capac-
`ity, those other devices may instead be used to perform these
`operations. Further in the alternative, while the described
`examples of the methods of this invention are realized by
`way of computer program routines executable by program-
`mable logic, it is also contemplated that these operations
`could be realized by custom hardwired logic, if desired.
`
`[0039] Referring to FIG. 5, measured values of signal-to-
`noise ratio (SNR) and signal-intersymbol interference-noise
`ratio (SINR) are received by LAN adapter 20 in process 30.
`As noted above, these parameters are estimated during
`reception of packet preamble PR, from another network
`element in the wireless LAN. Each communications chan-
`nel, being bidirectional, is evaluated by the network ele-
`ments on that channel, and the network elements may
`communicate their measurements of channel behavior to one
`another. In the case of the SNR and SINR measurements, the
`various network elements in the network are able to deter-
`
`Sonos Ex. 1010, p. 10
` Sonos v. Google
` IPR2021-00964
`
`

`

`US 2002/0122413 Al
`
`Sep. 5. 2002
`
`4
`
`mine these basic performance parameters by analysis of the
`known contents of packet preamble PR, as is known in the
`art. As known in the art, the SNR relates to the signal
`strength and noise within a given channel, while the SINR
`relates to the intersymbol interference experienced, for
`example because of multipath interference. These values
`thus are not intended to account for the presence of other
`interfering transmissions within the network range.
`
`[0040] Once the SNR and SINR values for the current
`channel are known, LAN adapter 20 calculates an expected
`packet error rate p, in process 32. Methods for the statistical
`determination of the probability that a given packet will fail,
`for a given set of SNR and SINR values, are known in the
`art. As noted above, however, this expected packet error rate
`pe will not fully appreciate the effects of interfering trans-
`missions, such as may be present due to Bluetooth or other
`sources.
`
`In process 34, IAN adapter 20 estimates the true
`[0041]
`packet error rate p, including the effects of interfering
`transmissions. This derivation is preferably performed for a
`channel by analyzing the number of packets that it transmits
`and that are not safely received, relative to the total number
`of packets transmitted over that channel. The criteria used to
`determine successful transmission is preferably based on the
`CRC checksum that is transmitted with the payload of each
`packet, and that the receiving network element, such as
`access point 10 for LAN adapter 20, checks against the
`payload portion PL of the packet itself. The overall results
`of the fraction of those packets that were received safely can
`be derived from acknowledgement messages from access
`point 10. Alternatively, the network elements in the wireless
`I,AN may communicate the success ratio for each channel
`by way of control packets. In any event, whether LAN
`adapter 20 calculates the packet error itself from acknowl-
`edge messages, or receives a packet error rate value from
`access point 10 or elsewhere in the network, the actual
`packet error rate p is determined in process 36.
`
`In decision 36, LAN adapter 20 determines
`[0042]
`whether the actual packet error rate p exceeds the expected
`packet error rate pe by more than a threshold amount e. If no
`interfering transmissions are present, the actual packet error
`rate p will vary about the expected error rate pe, either due
`to the approximation of the expected error rate p, or because
`of statistical variations. As such, threshold c is preferably set
`high enough so that, when exceeded, one can be confident
`that interfering transmissions are actually present in the
`network environment. If threshold e is not exceeded (deci-
`sion 36 is NO), control returns to process 30 to continue the
`monitoring of the packet error rates. If interfering transmis-
`sions appear to be present, based on the comparison of the
`expected error rate p, and the actual packet error rate p, LAN
`adapter 20 then detects the presence of an interferer, such as
`Bluetooth, that was not anticipated by the SNR and SINR,
`in process 40.
`
`It is contemplated that the process of FIG. 5 may
`[0043]
`he used to determine whether interfering transmissions are
`present in the network, in conjunction with a wide variety of
`optimization processes, including process 40 according to
`the preferred embodiment of the invention as will be
`described in detail below. Alternatively, once interference
`has been detected using the method of FIG. 5 according to
`the preferred embodiment of the invention, other conven-
`
`tional methods for avoiding the effects of interference may
`also be used, such methods including dynamic channel
`selection, conventional fragmentation, and the like.
`
`[0044] Referring now to FIG. 6, packet length optimiza-
`tion process 40 according to the preferred embodiment of
`the invention will now be described in detail. While process
`40 is preferably used in combination with the interference
`detection process of FIG. 5 discussed above, it is also
`contemplated that process 40 may be used with other means
`of detecting interference, or alternatively may be uncondi-
`tionally used upon the establishment of each and every
`communications channel.
`
`In process 42, learning constant II is initialized. As
`[0045]
`will become apparent below, learning constant it is a con-
`stant that determines the rate of change of packet length tp
`for a given calculated data rate change. The value of learning
`constant 14 is preferably determined empirically for the
`expected wireless LAN environment, considering that a
`value of learning constant ,t,/ that is too high will result in
`oscillation, while a value that is too low will delay optimi-
`zation. According to this example, a useful value of learning
`constant ,tt is about 1.275. In process 44, other parameters in
`optimization process 40 are initialized, including the appro-
`priate values of header and preamble time ti, and interpacket
`spacing to. In process 46, an initial packet length tpk and a
`prior initial packet length ti„k_i are set; in addition, a prior
`rate measure Fk_i may be set (to any arbitrary value,
`including zero).
`
`In process 48, a current packet success rate q'(tpk)
`[0046]
`for initial packet length tp ,, is determined. Measurement
`process 48 may be performed by the transmission of actual
`message packets at the current packet length tp,k and the
`calculation of an actual packet success rate based on check-
`sum results at the receiver, as described above relative to
`process 34 (FIG. 5). Alternatively, a success rate function
`q'(tp) may be known or assumed, in which case the packet
`success rate q'(tpk) is determined by applying the current
`packet length to , to this function.
`
`In the adaptive algorithm of optimization process
`[0047]
`40, a rate measure Fk is next derived, for the current packet
`length tp,k. This rate measure Fk operates as a fitness func-
`tion, in the adaptive algorithm sense, by way of which the
`packet length tp,,o, for the next iteration is determined.
`According to this preferred embodiment of the invention, an
`example of rate measure Fk is:
`
`-F5 = V(r,,k)x
`
`(to - tH )
`(tm + to)
`
`[0048] As evident from this definition of rate measure Fk,
`the parameters involved in the calculation of process 50
`includes the success rate (estimated or actual) for the current
`packet length tp,k, and a ratio of the actual data portion of the
`packet relative to the entire packet length, including its
`interpacket spacing. Other factors may also be used in this
`measure, including maximum data rate and the like, as
`desired
`
`In process 52, LAN adapter 20 derives difference
`[0049]
`value A based on the difference between the current and
`previous rate measures F, as follows:
`
`Sonos Ex. 1010, p. 11
` Sonos v. Google
` IPR2021-00964
`
`

`

`US 2002/0122413 Al
`
`Sep. 5, 2002
`
`5
`
`and
`[0050] The negative applied to rate measure
`difference value A reflects the use of these values as negative
`feedback in the adjustment of the packet length tp in process
`54. In process 54, the next packet length tp.„, is generated
`based on the difference A in the rate measures F, with the
`learning factor ft, as follows:
`
`tp0-1=tp.,,tuA
`[0051] Decision 55 is then performed, to determine if
`additional adjustment of the packet length by the execution
`of another iteration of the loop is necessary. The decision
`criterion used in decision 55 may be a convergence criterion,
`by way of which the difference between next packet length
`tp,k„ and prior packet length tp k is measured against a
`convergence threshold; alternatively, optimization process
`40 may be performed for a preselected number of iterations,
`in which case decision 55 would simply interrogate a loop
`counter. In any event, if the appropriate convergence crite-
`rion has not yet been met (decision 55 is YES), index k is
`incremented in process 56 so that the next packet length
`becomes the current packet length for purposes of this
`method, and control passes to process 48 for measurement
`or determina