Detailed Action
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 9/10/2025 has been entered.
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
Applicant's arguments filed 9/10/2025 have been fully considered but they are not persuasive.
The examiner has reviewed the Applicant’s arguments submitted on 9/10/2025 in their entirety (Pages 9-14). The Applicant states (Page 10) “ ... Englund teaches selecting between congestion windows sizes, but doe not each a rate of adjustment of a congestion window size ...”
Applicant’s argument is not persuasive because it is not commensurate with the scope of the claim. The claim requires “adjust[ing] the congestion window size ... to adjust the rate of packet transmission,” not “adjust[ing] the rate of adjustment of a congestion window size” as argued. England discloses selecting an initial permissible congestion window size during the slow start phase ([0042]), which directly results in a higher packet transmission rat. This meets the claim requirement of adjusting the congestion window size to adjust the transmission rate. Accordingly, Applicant’s attempt to reframe the claim to require disclosure of a frate of adjustment of the congestion window is misplaced.
Moreover, Mindler (US 2021/0160183) expressly discloses “increasing the rate at which cwnd increases during the slow start phase” and increasing the rate ... albeit at a much slower rate during the congestion avoidance phase “ While this “rate at which cwnd increases” is not required by the claims, it nevertheless confirms that such rate-base behavior of the congestion window parameter is ell-recognized in the art and directly associated with the operation of TCP slow-start and congestion-avoidance phases.
The Examiner respectfully reminds the Applicant the prior art rejection (Agarwal, Callaghan), discussed on Page 12 does not exist in the Final Office Action furnished on 6/10/2025.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 6, 10, 13, 17, 19, 25, 29, 32 are rejected under 35 USC 103 as being unpatentable over Englund (US 2016/0380898) in view of Dong (US 2021/0051102) and in further view of Allman, “RFC 5681”, September 2009
Regarding claim 1, Englund discloses an apparatus comprising:
a network interface device comprising (Englund; TCP nodes inherently are network interface devices; see e.g. [0042}):
a network interface (Englund; TCP nodes inherently comprise conventional network interfaces to communicate with other TCP device; see e.g. [0042}and
circuitry to (Englund; Circuitry is inherently present to provide operations):
adjust a rate of packet transmissions (Englund;
see e.g. [0042] “The TCP sending node 1 is provided with information so that it can select a permissible initial permissible congestion window size for a slow start mechanism when sending TCP data. This allows the slow start mechanism to select an initial permissible congestion window size that allows the initial data bit rate to be higher than a standard initial window size if conditions allow this. This is illustrated in FIG. 3, where an initial permissible congestion window size is selected by the TCP sending node such that the initial bit rate is substantially higher than initial bit rate of a standard TCP slow start mechanism, as shown in FIG. 1. An advantage of this is that the TCP receiving node starts receiving data at a higher bit rate and so the end user's QoE is not affected so much by the slow start mechanism. Note that references to the bit rate refer to the average bit rate; the bit rate at any given time fluctuates around an average value”
see e.g. [0025] “ ... the TCP sending node further comprises a database containing data mapping parameter values to corresponding permissible window sizes)”
:
the adjust the rate of packet transmission based on the phase of operation and congestion metrics comprises (Englund; see e.g. [0042]):
during a Transmission Control Protocol (TCP) slow start phase of operation comprising no indicated congestion, adjust a congestion window size by a first level of change to adjust the rate of packet transmission (Englund; England discloses a slow start mechanism, which by definition begins in the absence of indicated congestion. England further teaches selecting an initial permissible congestion window size such that the transmission rate is increased. Thus, England teaches adjusting the congestion window size by a first level of change to adjust the rate of packet transmission as recited.
see e.g. Abstract “A method and apparatus for controlling a Transmission Control Protocol (TCP) window size for data sent from a TCP sending node via a mobile network. The TCP sending node receives a TCP request message from a remote node requesting a TCP data stream. The TCP request message includes a parameter relating to conditions in the mobile network and/or the presence of a TCP proxy node. The TCP sending node uses the parameter to determine an initial permissible window size for a TCP slow start mechanism. It can then send a TCP data stream towards a TCP receiving node using the slow start mechanism starting with the initial permissible window size. This allows the slow start phase to start with a higher initial window size”
see e.g. [0042] “The TCP sending node 1 is provided with information so that it can select a permissible initial permissible congestion window size for a slow start mechanism when sending TCP data. This allows the slow start mechanism to select an initial permissible congestion window size that allows the initial data bit rate to be higher than a standard initial window size if conditions allow this. This is illustrated in FIG. 3, where an initial permissible congestion window size is selected by the TCP sending node such that the initial bit rate is substantially higher than initial bit rate of a standard TCP slow start mechanism, as shown in FIG. 1. An advantage of this is that the TCP receiving node starts receiving data at a higher bit rate and so the end user's QoE is not affected so much by the slow start mechanism. Note that references to the bit rate refer to the average bit rate; the bit rate at any given time fluctuates around an average value”see e.g. [0025]) “,
during a congestion avoidance phase of operation, based on indication of congestion, adjust the congestion window size by a second level of change to adjust the rate of packet transmission transmissions (Englund; While England explicitly details the slow start phase, one of ordinary skill in the art would recognize that TCP inherently transitions from slow into congestion avoidance once congestion is indicated. In congestion avoidance, the congestion window size is adjusted by a second, more conservative level of change (liner growth rather than the more aggressive initial change of slow start). Since the claim requires that , in congestion avoidance, the congestion window is adjusted by a second level of change to adjust the rate of packet transmission, and since transmission rate is directly proportional to congestion window size , England’s’ disclosure of slow start, in combination with the inherent transition to congestion avoidance once congestion is indicated, teaches the recited limitation.
see e.g. [0042] ,see e.g. Abstract)
the first level is higher than the second level (Englund; England explicitly teaches that during the slow start phase, the sending node selects an initial permissible congestion window size that produces a substantially higher initial bit rate than a standard start up condition. The slow start operation therefore applies a comparatively greater intial change to the congestion window (a higher level) than the later, more conservative adjustments that occur once congestion becomes indicated. Because the transition from slow-start to congestion avoidance inherently entails moving from a more aggressive growth behavior to a less aggressive one, England implicitly teaches that the first level of change (applied during slow start) is higher than the second level of change (applied once congestion is indicated
see e.g. [0025], [0042], Fig. 3), and
the first level of change is to cause a faster adjustment to the congestion window size than the adjustment to the congestion window size caused by the second level of change (Englund; England describes that by starting with a higher initial congestion window size , the sending node achieves a higher initial bit rate and thus a faster rise in transmission throughout at least the beginning of the connection. This faster increase in rate corresponds to a faster adjustment of the congestion window size -- the first level of change. Once the congestion becomes indicated, the protocol inherently transitions to congestion avoidance, in which the congestion window is adjusted more gradually to maintain stability and avoid further congestion. Accordingly, England implicitly teaches that the first level of changes (applied during the no -congestion slow -start phase) causes a faster adjustment to the congestion window size than the second level of change (applied during the congestion avoidance phase;
see e.g. Abstract, [0025], [0042]).
However Englund does not expressly disclose:
based on a phase of operation and congestion metrics comprising queue depth at one or more intermediate switches
However in analogous art Dong discloses:
based on a phase of operation and congestion metrics comprising queue depth at one or more intermediate switches (Dong;
see e.g. [0046] “a congestion flag may be set for a data packet, and the data packet that carries the congestion flag indicates connection congestion. The congestion flag is marked at a third layer of a network, that is, an IP layer. That is, the congestion flag is marked in an IP header of the data packet. The congestion flag indicates that congestion occurs in the network. For example, the congestion flag may be congestion experienced (CE). For example, the CE is 11 that is set in a header of an IP data packet. A data packet that does not carry a congestion flag indicates that congestion does not occur over the connection. It should be noted that a queue depth of the switching node is within a normal range of a queue depth, and the congestion flag setting probability is positively correlated with the queue depth. The queue depth of the switching node is a quantity of to-be-sent data packets. In an example, the normal range of the queue depth is related to one or more of the following parameters: a bandwidth-delay product, a maximum value of the normal range of the queue depth, a minimum value of the normal range of the queue depth, an increase factor, a maximum congestion flag setting probability, a third value corresponding to a first packet, and a fifth value corresponding to a second packet received by the transmit end in a Slow Start state. The Slow Start state is a state in a period of time when the connection is initially established. The slow start state starts when the connection is initially established, and ends when the transmit end receives the first packet for the first time”
see e.g. Fig. 2 [0041])
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
As evidence of the rationale above, Allman discloses:
the first level of change is higher than the second level of change (Allman;
see e.g. Page 6 “During slow start, a TCP increments cwnd by at most SMSS bytes for each ACK received that cumulatively acknowledges new data. ... Note that during congestion window avoidance, cwnd must not be increased by more than SMSS bytes per RTT”
The evidence confirms that in slow start the congestion window is adjusted for every acknowledgement (multiple increments per RTT), whereas in congestion avoidance it is adjusted at most once per RTT. Accordingly, the first level of change—corresponding to the congestion avoidance behavior. This cohobates the rationale that Englunds’s disclosure of a slow-start mechanism in combination with the inherent transition to congestion avoidance when congestion is indicated as explicitly taught by RFC 5681, teaches the claimed first and second levels.)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Allman’s congestion scheme. The motivation being the combined solution provides for implementing a known technique resulting in increased efficiencies of managing and controlling of network traffic.
Regarding claim 6, Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, wherein the rate of packet transmissions is based on the congestion window size (Englund;
See e.g. [0042] “The TCP sending node 1 is provided with information so that it can select a permissible initial permissible congestion window size for a slow start mechanism when sending TCP data. This allows the slow start mechanism to select an initial permissible congestion window size that allows the initial data bit rate to be higher than a standard initial window size if conditions allow this. This is illustrated in FIG. 3, where an initial permissible congestion window size is selected by the TCP sending node such that the initial bit rate is substantially higher than initial bit rate of a standard TCP slow start mechanism, as shown in FIG. 1. An advantage of this is that the TCP receiving node starts receiving data at a higher bit rate and so the end user's QoE is not affected so much by the slow start mechanism. Note that references to the bit rate refer to the average bit rate; the bit rate at any given time fluctuates around an average value” see e.g. [0044]
Dong;
See e.g. [0040] The congestion control method in the embodiments of this application is mainly applied to the transmit end over the connection. The connection includes the transmit end, the switching node, and the receive end. The transmit end may send a data packet to the switching node, and the data packet enters a queue in the switching node. The switching node may be configured to set (that is, mark) a congestion flag for the data packet. The switching node sequentially sends data packets to the receive end based on a sequence of the data packets in the queue. The receive end sends a packet to the transmit end based on a received data packet. The transmit end adjusts the value of the congestion window based on the packet.).
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Regarding claim 10, Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, wherein the network interface device comprises one or more of: network interface controller (NIC), SmartNIC, router, forwarding element, infrastructure processing unit (IPU), or data processing unit (DPU)( Englund teaches TCP sending nodes which are equivalent to forwarding elements;
See e.g. [0004] “... TCP sending node ...”).
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Regarding claim 13, claim 13 comprises the same and/or similar subject matter as claim 1 and is considered an obvious variation; therefore it is rejected under the same rationale.
Regarding claim 17, claim 17 comprises the same and/or similar subject matter as claim 6 and is considered an obvious variation; therefore it is rejected under the same rationale.
Regarding claim 19, claim 19 comprises the same and/or similar subject matter as claim 1 and is considered an obvious variation; therefore it is rejected under the same rationale.
Regarding claim 25, Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, wherein:
the congestion metrics include a queue depth level and percentage of utilized bandwidth (The combined solution per Dong;
see e.g. [0046] “a congestion flag may be set for a data packet, and the data packet that carries the congestion flag indicates connection congestion. The congestion flag is marked at a third layer of a network, that is, an IP layer. That is, the congestion flag is marked in an IP header of the data packet. The congestion flag indicates that congestion occurs in the network. For example, the congestion flag may be congestion experienced (CE). For example, the CE is 11 that is set in a header of an IP data packet. A data packet that does not carry a congestion flag indicates that congestion does not occur over the connection. It should be noted that a queue depth of the switching node is within a normal range of a queue depth, and the congestion flag setting probability is positively correlated with the queue depth. The queue depth of the switching node is a quantity of to-be-sent data packets. In an example, the normal range of the queue depth is related to one or more of the following parameters: a bandwidth-delay product, a maximum value of the normal range of the queue depth, a minimum value of the normal range of the queue depth, an increase factor, a maximum congestion flag setting probability, a third value corresponding to a first packet, and a fifth value corresponding to a second packet received by the transmit end in a Slow Start state. The Slow Start state is a state in a period of time when the connection is initially established. The slow start state starts when the connection is initially established, and ends when the transmit end receives the first packet for the first time”)
see e.g. Fig. 2 [0041] ),
during the congestion avoidance phase of operation, the second level of change comprises a multiplier-based change in the congestion window size (The Examiner notes this limitation is taught by Applicant Admitted Prior Art from the Non Final Office Action of 11/21/2024 where dependent claim 2 was not adequately traversed. Dependent claim 2 states “,,, where the circuitry is to adjust the rate of packet transmissions by multiplicative decrease or increase based on a number of inflight bytes” This behavior contemplates a multiplier based control function dependent on in-flight data (i.e., bandwidth delay product), which corresponds to “an amount of data in transit” in the present claim. A person of ordinary skill in the art would have found it obvious to combine Englunds TCP window adjustment with Dong’s congestion feedback inputs, and apply the known multiplier -based logic from AAPA to achieve dynamic control over the congestion window. The motivation being a multiplier based second level of change, based on queue depth, bandwidth utilization, and amount of inflight data and with the congestion window size controlling the transmit rate of packets.)
the multiplier-based change in the congestion window size is based on the queue depth level and the percentage of utilized bandwidth and an amount of data in transit (The Examiner notes this limitation is taught by Applicant Admitted Prior Art from the Non Final Office Action of 11/21/2024 where dependent claim 2 was not adequately traversed. Dependent claim 2 states “,,, where the circuitry is to adjust the rate of packet transmissions by multiplicative decrease or increase based on a number of inflight bytes” This behavior contemplates a multiplier based control function dependent on in-flight data (i.e., bandwidth delay product), which corresponds to “an amount of data in transit” in the present claim. A person of ordinary skill in the art would have found it obvious to combine Englund’s TCP window adjustment with Dong’s congestion feedback inputs, and apply the known multiplier -based logic from AAPA to achieve dynamic control over the congestion window. The motivation being a multiplier based second level of change, based on queue depth, bandwidth utilization, and amount of inflight data and with the congestion window size controlling the transmit rate of packets),
the amount of data in transit comprises a Bandwidth Delay Product (BDP) (The combined solution per AAPA; as detailed above “,,, where the circuitry is to adjust the rate of packet transmissions by multiplicative decrease or increase based on a number of inflight bytes”)
The Examiner notes the number of in-flight bytes is equivalent to the bandwidth delay product) , and
the congestion window size is to control a transmit rate of packets (The combined solution per Englund and Dong as the congestion window mechanism taught by Englund (see e.g. [0042]) is a pacing mechanism that controls the amount of data injected into the network resulting in the rate at which packets arrive at downstream queues via Dong (see. e.g. [0046], Fig.2)..
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Regarding claim 29, claim 29 comprises the same and/or similar subject matter as claim 25 and is considered an obvious variation; therefore it is rejected under the same rationale;
Regarding claim 32, claim 32 comprises the same and/or similar subject matter and is considered an obvious variation; therefore it is rejected under the same rationale;
Claims 24, 28, and 31 are rejected under 35 USC 103 as unpatentable over Englund in view of Dong and in further view of Allman and in further view of Li et al, “HPCC: high precision congestion control”, September 2019
Regarding claim 24, Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, wherein:
during the TCP slow start phase of operation, the first level of change comprises a multiplier-based increase in the congestion window size and the first level of change is based on the reported utilization level value (The combined solution per Englund and Dong; see e.g. [0042], Dong see e.g. [0046]), and
the congestion window size is to control a transmit rate of packets to the queue (The combined solution per Englund and Dong; see e.g. [0042], Dong see e.g. [0046]).
Englund in view of Dong and in further view of Allman does not expressly disclose:
the congestion metrics include a reported utilization level value consistent with High Precision Congestion Control (HPCC) protocol,
However in analogous art Li discloses:
the congestion metrics include a reported utilization level value consistent with High Precision Congestion Control (HPCC) protocol (Li;
see e.g. Abstract;
see e.g. Section 5.1: Evaluation setup)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Li’s HPCC telemetry scheme. The motivation being the combined solution provides for implementing a known technique resulting in increased efficiencies of packet transmission.
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Regarding claim 28, claim 28 comprises the same and similar subject matter as claim 24 and is considered an obvious variation; therefore it is rejected under the same rationale.
Regarding claim 31, claim 31 comprises the same and similar subject matter as claim 24 and is considered an obvious variation; therefore it is rejected under the same rationale.
Claims 22-23, 27, and 30 is rejected under 35 USC 103 as being unpatentable over Englund in view of Dong and in further view of Allman (“RFC 5861”) and in further view of Attar (US 9,001,663) and in further view of Bensley (“RFC 8257”).
Regarding claim 22, Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, wherein:
the congestion window size is to control a transmit rate of packets to the queue (The combined solution per Englund and Dong as the congestion window mechanism taught by Englund (see e.g. [0042]) is a pacing mechanism that controls the amount of data injected into the network resulting in the rate at which packets arrive at downstream queues via Dong (see. e.g. [0046], Fig.2).
during the congestion avoidance phase of operation, the second level of change comprises an addition-based change in the congestion window size (The combined solution per Allman teaches .an addition-based change in the congestion window size during the congestion avoidance phase;
see e.g. Page 6, Section 3.1”... During congestion avoidance, cwnd is incremented by roughly 1 full sized segment per round-trip time (RTT) ...”)
see e.g. Page 3, Section 2. “... Congestion Window (cwnd) ...”)
The combined solution does not expressly disclose:
the addition-based change in the congestion window size is based on a larger of: (1) a ratio of weightings of queue depth values received in the congestion metrics and a queue depth level based on an Explicit Congestion Notification (ECN) marking and (2) a fraction above a target queue depth and consistent with Data Center TCP (DCTCP), and
However in analogous art Attar discloses:
the addition-based change in the congestion window size is based on (2) a fraction above a target queue depth and consistent with Data Center TCP (DCTCP) (Attar;
see e.g. Column 7, Lines 19 – 24 “The DCTCP senders start reducing their window size as soon as the queue size exceeds K. The DCTCP algorithm thus maintains low queue size, while ensuring high throughput . When the value of alpha is near 1(high congestion), the DCTCP algorithm reduces its window by half, as in TCP”
see e.g. Column 8, Lines 50 - 52 “In act 608, a running estimate of the fraction of marked packets is updated by sender 300 using Equation (1) above ...”)
see e.g. Column 6, Lines 55-61),
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Attar’s DCTCP scheme. The motivation the implantation of a known technique results in increased efficiencies of packet transmissions.
Englund in view of Dong and in further view of Allman (“RFC 5861”) and in further view of Attar does not expressly disclose:
(1) a ratio of weightings of queue depth values received in the congestion metrics and a queue depth level based on an Explicit Congestion Notification (ECN) marking
However in analogous art Bensley discloses:
(1) a ratio of weightings of queue depth values received in the congestion metrics and a queue depth level based on an Explicit Congestion Notification (ECN) marking (Bensley, see e.g. Section 3.3;
The Examiner notes although G is labelled an estimation gain, it function servers as a weight o that balances past vs. current ECN based congestion indications. This weighting of Q depth values through g—together with F as the ration of ECN marked packets constitutes the claimed ration of weightings of Q-depth values and an ECN based Q-depth level.)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Bensley’s ECN scheme. The motivation being the combined solution provides for incorporating a known technique resulting in increased efficiencies of packet transmission. More importantly it would have been obvious to one of ordinary skill in the art to select the larger of items (1) and (20 to determine an addition based change in the congestion window size because doing so allows the system to respond more aggressively to whichever congestion signal indicates higher stress in the network. Specifically, when ECN marking may underrepresent instantaneous queue buildup (due to marking thresholds or delay), using the DCTCP based fractional queue depth offers finer granularity,. Conversely, when ECN feedback is more timely or accurate, it may capture congestion better than raw queue length. Taking the larger of the two ensures the system errs on the side of reducing transmission more assertively, thus improving stability, reducing packet loss, and ensuring high throughput and low latency across diverse network conditions. Such a hybrid logic would be a predictable combination of well understood congestion metrics to improve network responsiveness.
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Regarding claim 23, Englund in view of Dong and in further view of Allman and in further view of Attar and in further view of Bensley disclose the apparatus of claim22, wherein the circuitry is to adjust the rate of packet transmission based on an expiration of round trip time (RTT) (Englund discloses a TCP transmission control scheme employing a congestion window (cwnd) to manage the rate of data transmission from a sender node (see e.g. [0042). Englund teaches the window size is adjusted based on transmission phase and network conditions, thereby controlling the rate at which packets are sent. Allman provides further details on standard TCP behavior and specifically states (Page, 6, Section 3.1 “,,, During congestion avoidance, cwnd is incremented by roughly 1 full size segment per round-trip time (RTT) Allman also disclose that the sender should increment CWND per equation 2 (Page 6, Section 3.1). This confirms that cwnd based adjustments in transmission rate once per RTT, meaning the rate of the packet transmission is inherently adjusted based on the expiration of the RTT)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Regarding claim 27, claim 27 comprises the same and similar subject matter as claim 22 and is considered an obvious variation; therefore it is rejected under the same rationale.
Regarding claim 30, claim 30 comprises the same and similar subject matter as claim 22 and is considered an obvious variation; therefore it is rejected under the same rationale.
Claims 3, 8, 18 and 26 are rejected under 35 USC 103 as being unpatentable over Englund in view of Dong and in further view of Allman and in further view of Lee (US 2020/0280518)
Regarding claim 3, Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, but does not expressly disclose wherein the circuitry is to adjust the rate of packet transmissions by additive decrease or increase based on congestion metric from the one or more intermediate switches at one or more intermediate switches.
However in analogous art Lee discloses:
wherein the circuitry is to adjust the rate of packet transmissions by additive decrease or increase based on congestion metric from the one or more intermediate switches at one or more intermediate switches (Lee, Lee teaches a metric known as drain rate which is related to queues and queue depth which can utilized to effect the rate of packet transmission;
see e.g. [0035] “For example, congestion monitor 272 can track the following information for a flow to predict whether the packet will reach a congested queue and determine a pause time to be applied by network element 250-0 for other packets in the same flow. The information associated with the relationship described with respect to FIG. 3 can be stored in a state table associated with a match table. For example, the following parameters can be tracked”
PNG
media_image1.png
136
472
media_image1.png
Greyscale
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Lee’s adaptive transmission scheme. The motivation being the combined solution provides for implementing a known technique resulting in increased efficiencies of controlling packet flows.
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Regarding claim 8, Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, wherein the congestion metrics include link utilization (The combined solution per Dong;
See e.g. [0074] “In addition, in this embodiment of this application, when the stable queue depth is not 0, that is, when it is ensured that there is always a data packet in the queue, comparatively high network bandwidth utilization can also be ensured. This can achieve a balance between a low delay and high network bandwidth utilization in a data transmission process”) however Englund in view of Dong and in further view of Allman does not expressly disclose:
wherein the congestion metrics include a number of flows that provides packets to a congested queue;
However in analogous art Lee discloses:
wherein the congestion metrics include a number of flows that provides packets to a congested queue (Lee;
see e.g [0028] “For example, elephants flows (e.g., large flows), mice flows (e.g., small flows) sharing a same congested port and queue can experience tail drops or ECN marking. By use of an accurate specification of pause time, occurrences of head of line blocking for flows sharing the same link (or queue) from an upstream network element to the congested network element can be reduced compared to use of Priority-based Flow Control (PFC) or pause-based Xon/Xoff”
see e.g. [0046] In some examples, a sender pause time could be set not to completely drain the congested queue but to drain only down to the target (or reference) queue depth used by congestion control schemes. For example, ECN marking algorithms (such as random early detection (RED) or Proportional-Integral (PI)) can include target queue depth as part of parameter configurations. If the SQP pause time is set too long and drains the queue below the target queue depth, the link utilization may go below 100%, hurting application throughput)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Lee’s teachings. The motivation being the combined solution provides for implementing a known technique resulting in increased network transmittal efficiencies.
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Regarding claim 18, claim 18 comprises the same and/or similar subject matters as claim 8 and is considered an obvious variation; therefore it is rejected under the same rationale.
Regarding claim 26, Englund in view of Dong and in further view of Allman discloses the apparatus of claim 1 Englund in view of Dong and in further view of Allman does not expressly disclose wherein:
the circuitry is to adjust the congestion window size based on a number of flows contributing to congestion in the queue.
However in analogous art Lee discloses:
the circuitry is to adjust the congestion window size based on a number of flows contributing to congestion in the queue (Lee;
see e.g [0028] “For example, elephants flows (e.g., large flows), mice flows (e.g., small flows) sharing a same congested port and queue can experience tail drops or ECN marking. By use of an accurate specification of pause time, occurrences of head of line blocking for flows sharing the same link (or queue) from an upstream network element to the congested network element can be reduced compared to use of Priority-based Flow Control (PFC) or pause-based Xon/Xoff”
see e.g. [0046] In some examples, a sender pause time could be set not to completely drain the congested queue but to drain only down to the target (or reference) queue depth used by congestion control schemes. For example, ECN marking algorithms (such as random early detection (RED) or Proportional-Integral (PI)) can include target queue depth as part of parameter configurations. If the SQP pause time is set too long and drains the queue below the target queue depth, the link utilization may go below 100%, hurting application throughput)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Lee’s teachings. The motivation being the combined solution provides for implementing a known technique resulting in increased network transmittal efficiencies.
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Claim 5 is rejected under 35 USC 103 as being unpatentable over Englund in view of Dong and in further view of Allman and in further view of Tang (US 2018/0013677)
Regarding claim 5, Englund in view of Dong and in further view of Allman disclose the apparatus of claim1, but does not expressly disclose wherein the circuitry is to apply Data Center TCP (DCTCP) congestion control based on unavailability of queue depth data from the congestion metrics
However in analogous art Tang discloses:
wherein the circuitry is to apply Data Center TCP (DCTCP) congestion control based on unavailability of queue depth data from the congestion metrics (Tang; Tang discloses the utilization of DCTCP prior to a link being fully utilized to mitigate latency hence providing one of ordinary skill in the art to provide congestion control without queue depth information;
see e.g. [0006] “... improve the efficiency through minimizing latency is to signal the endpoint device about a congestion before a link is fully utilized ... High Bandwidth Ultra Low Latency (HULL) architecture which is an extension of a Data Centre Transmission Control Protocol (DCTCP) ...”)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Tang’s conventional use of DCTCP. The motivation being the combined solution provides for implementing a known technique resulting in increased efficiencies of packet transmission.
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Claim 11 is rejected under 35 USC 103 as being unpatentable over Englund in view of Dong and in further view of Allman and in further view of Attar (US 9,001,663)
Regarding claim 11, Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, but does not expressly disclose comprising a computing system coupled to the network interface device, wherein the computing system is to configure the congestion window size of the network interface device to adjust a rate of packet transmissions to the queue based on Data Center TCP (DCTCP) congestion control.
However in analogous art Attar discloses:
comprising a computing system coupled to the network interface device, wherein the computing system is to configure the congestion window size of the network interface device to adjust a rate of packet transmissions to the queue based on Data Center TCP (DCTCP) congestion control (Attar;
see e.g. Column 7, Lines 19 – 24 “The DCTCP senders start reducing their window size as soon as the queue size exceeds K. The DCTCP algorithm thus maintains low queue size, while ensuring high throughput . When the value of alpha is near 1(high congestion), the DCTCP algorithm reduces its window by half, as in TCP”
see e.g. Column 8, Lines 50 - 52 “In act 608, a running estimate of the fraction of marked packets is updated by sender 300 using Equation (1) above ...”)
see e.g. Column 6, Lines 55-61),
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Attar’s DCTCP scheme. The motivation benig the implementation of a known technique results in increased efficiencies of packet transmissions.
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Claim 12 is rejected under 35 USC 103 as being unpatentable over Englund in view of Dong and in further view of Allman and in further view of Bergeron (US 2016/0057040)
Regarding claim 12, Although Englund in view of Dong and in further view of Allman disclose the apparatus of claim 1, and Englund view of Dong and in further view of Allman discloses the congestion in the queue as indicated by congestion metrics (see e.g. Dong [0074]) Englund in view of Dong and in further view of Allman does not address the frequency based on a number of flows and therefore does not expressly disclose comprising one or more switches to provide the congestion metrics at a frequency based on a number of flows contributing to congestion in the queue as indicated by the congestion metrics.
However in analogous art Bergeron discloses:
one or more switches to provide the congestion metrics at a frequency based on a number of flows contributing to congestion (Bergeron;
see e.g. [0025] “... frequency of metric generation, a number of packets per flow to per generated ... a number of flows ...”)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Bergeron’s scheme. The motivation being the implementation of a known technique results in increased efficiencies of packet transmissions.
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
Claims 1, 13, and 19 are rejected under 35 USC 103 as being unpatentable over Englund (US 2016/0380898) in view of Dong (US 2021/0051102) and in further view of Mindler (US 2021/0160183)
Regarding claim 1, Englund discloses an apparatus comprising:
a network interface device comprising (Englund; TCP nodes inherently are network interface devices; see e.g. [0042}):
a network interface (Englund; TCP nodes inherently comprise conventional network interfaces to communicate with other TCP device; see e.g. [0042}and
circuitry to (Englund; Circuitry is inherently present to provide operations):
adjust a rate of packet transmissions (Englund;
see e.g. [0042] “The TCP sending node 1 is provided with information so that it can select a permissible initial permissible congestion window size for a slow start mechanism when sending TCP data. This allows the slow start mechanism to select an initial permissible congestion window size that allows the initial data bit rate to be higher than a standard initial window size if conditions allow this. This is illustrated in FIG. 3, where an initial permissible congestion window size is selected by the TCP sending node such that the initial bit rate is substantially higher than initial bit rate of a standard TCP slow start mechanism, as shown in FIG. 1. An advantage of this is that the TCP receiving node starts receiving data at a higher bit rate and so the end user's QoE is not affected so much by the slow start mechanism. Note that references to the bit rate refer to the average bit rate; the bit rate at any given time fluctuates around an average value”
see e.g. [0025] “ ... the TCP sending node further comprises a database containing data mapping parameter values to corresponding permissible window sizes)”
:
the adjust the rate of packet transmission based on the phase of operation and congestion metrics comprises (Englund; see e.g. [0042]):
during a Transmission Control Protocol (TCP) slow start phase of operation comprising no indicated congestion, adjust a congestion window size by a first level of change to adjust the rate of packet transmission (Englund; England discloses a slow start mechanism, which by definition begins in the absence of indicated congestion. England further teaches selecting an initial permissible congestion window size such that the transmission rate is increased. Thus, England teaches adjusting the congestion window size by a first level of change to adjust the rate of packet transmission as recited.
see e.g. Abstract “A method and apparatus for controlling a Transmission Control Protocol (TCP) window size for data sent from a TCP sending node via a mobile network. The TCP sending node receives a TCP request message from a remote node requesting a TCP data stream. The TCP request message includes a parameter relating to conditions in the mobile network and/or the presence of a TCP proxy node. The TCP sending node uses the parameter to determine an initial permissible window size for a TCP slow start mechanism. It can then send a TCP data stream towards a TCP receiving node using the slow start mechanism starting with the initial permissible window size. This allows the slow start phase to start with a higher initial window size”
see e.g. [0042] “The TCP sending node 1 is provided with information so that it can select a permissible initial permissible congestion window size for a slow start mechanism when sending TCP data. This allows the slow start mechanism to select an initial permissible congestion window size that allows the initial data bit rate to be higher than a standard initial window size if conditions allow this. This is illustrated in FIG. 3, where an initial permissible congestion window size is selected by the TCP sending node such that the initial bit rate is substantially higher than initial bit rate of a standard TCP slow start mechanism, as shown in FIG. 1. An advantage of this is that the TCP receiving node starts receiving data at a higher bit rate and so the end user's QoE is not affected so much by the slow start mechanism. Note that references to the bit rate refer to the average bit rate; the bit rate at any given time fluctuates around an average value”see e.g. [0025]) “,
during a congestion avoidance phase of operation, based on indication of congestion, adjust the congestion window size by a second level of change to adjust the rate of packet transmission transmissions (Englund; While England explicitly details the slow start phase, one of ordinary skill in the art would recognize that TCP inherently transitions from slow into congestion avoidance once congestion is indicated. In congestion avoidance, the congestion window size is adjusted by a second, more conservative level of change (liner growth rather than the more aggressive initial change of slow start). Since the claim requires that , in congestion avoidance, the congestion window is adjusted by a second level of change to adjust the rate of packet transmission, and since transmission rate is directly proportional to congestion window size , England’s’ disclosure of slow start, in combination with the inherent transition to congestion avoidance once congestion is indicated, teaches the recited limitation.
see e.g. [0042] ,see e.g. Abstract)
the first level is higher than the second level (Englund; England explicitly teaches that during the slow start phase, the sending node selects an initial permissible congestion window size that produces a substantially higher initial bit rate than a standard start up condition. The slow start operation therefore applies a comparatively greater intial change to the congestion window (a higher level) than the later, more conservative adjustments that occur once congestion becomes indicated. Because the transition from slow-start to congestion avoidance inherently entails moving from a more aggressive growth behavior to a less aggressive one, England implicitly teaches that the first level of change (applied during slow start) is higher than the second level of change (applied once congestion is indicated
see e.g. [0025], [0042], Fig. 3), and
the first level of change is to cause a faster adjustment to the congestion window size than the adjustment to the congestion window size caused by the second level of change (Englund; England describes that by starting with a higher initial congestion window size , the sending node achieves a higher initial bit rate and thus a faster rise in transmission throughout at least the beginning of the connection. This faster increase in rate corresponds to a faster adjustment of the congestion window size -- the first level of change. Once the congestion becomes indicated, the protocol inherently transitions to congestion avoidance, in which the congestion window is adjusted more gradually to maintain stability and avoid further congestion. Accordingly, England implicitly teaches that the first level of changes (applied during the no -congestion slow -start phase) causes a faster adjustment to the congestion window size than the second level of change (applied during the congestion avoidance phase;
see e.g. Abstract, [0025], [0042]).
However Englund does not expressly disclose:
based on a phase of operation and congestion metrics comprising queue depth at one or more intermediate switches
However in analogous art Dong discloses:
based on a phase of operation and congestion metrics comprising queue depth at one or more intermediate switches (Dong;
see e.g. [0046] “a congestion flag may be set for a data packet, and the data packet that carries the congestion flag indicates connection congestion. The congestion flag is marked at a third layer of a network, that is, an IP layer. That is, the congestion flag is marked in an IP header of the data packet. The congestion flag indicates that congestion occurs in the network. For example, the congestion flag may be congestion experienced (CE). For example, the CE is 11 that is set in a header of an IP data packet. A data packet that does not carry a congestion flag indicates that congestion does not occur over the connection. It should be noted that a queue depth of the switching node is within a normal range of a queue depth, and the congestion flag setting probability is positively correlated with the queue depth. The queue depth of the switching node is a quantity of to-be-sent data packets. In an example, the normal range of the queue depth is related to one or more of the following parameters: a bandwidth-delay product, a maximum value of the normal range of the queue depth, a minimum value of the normal range of the queue depth, an increase factor, a maximum congestion flag setting probability, a third value corresponding to a first packet, and a fifth value corresponding to a second packet received by the transmit end in a Slow Start state. The Slow Start state is a state in a period of time when the connection is initially established. The slow start state starts when the connection is initially established, and ends when the transmit end receives the first packet for the first time”
see e.g. Fig. 2 [0041])
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Dong’s queue depth based congestion detection logic as a trigger for determining whether to remain in slow start or transition to congestion avoidance. The use of queue depth as a real-time congestion indicator provides a more accurate and granular signal of network load conditions, which would enhance the decision making process in Englund’s phase based congestion window control scheme.
As evidence of the rationale above, Mindler discloses:
the first level of change is higher than the second level of change, (Mindler; Mindler explicitly teaches that the congestion window parameter (cwnd) is increased by a full segment size during the TCP slow- start pahse and by only a fractional increment during the congestion-avoidance phase, resulting in “a much slower rate” of cwnd increase during congestion avoidance. Accordingly, Mndler explicitly teaches that the first level of change (slow-start) adjustment) is higher than the second level of change and causes a faster adjustment to congestion window size caused by the second level of change
see e.g. [0053] “P-GW 115 modifies one or more TCP congestion control send parameters to ramp up the data transmission rate to the UE 100 via the Next Generation RAN 330 (block 550). In some implementations, P-GW 115 may modify the TCP slow start and/or congestion window (CWND) parameters to ramp up the rate of transmission of data (i.e., quickly increase the data throughput)
sent from P-GW 115 to the UE 100 via the Next Generation RAN 330. P-GW 115 may modify the one or more TCP congestion control send parameters responsive to receipt of the network context change notification. The slow start phase of the TCP congestion control algorithm may be represented by the following pseudocode”
see e.g. [0056] P-GW 115 may modify the CWND parameter by increasing the value of segsize such that the CWND increments by the increased segsize value (i.e., CWND=CWND+segsize), thereby increasing the rate at which CWND increases during the slow start phase of TCP congestion control.
see e.g. [0057] IF P-GW 115 modifies the CWND parameter by increasing the value of segsize, then CWND, during the congestion avoidance phase, also increments by some percentage of the increased segsize value (i.e., CWND+segsize*segsize/CWND), thereby increasing the rate at which CWND increases (albeit at a much slower rate than during the slow start phase)
Therefore it would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate MIndler’s congestion scheme. The motivation being the combined solution provides for implementing a known technique resulting in increased efficiencies of managing and controlling of network traffic. Moreover, Mindler provides for a first, faster adjustment of cwnd during the slow-start phase and a slower adjustment of cwnd during the congestion avoidance phase.
Regarding claim 13, claim 13 comprises the same and/or similar subject matter as claim 1, and is considered an obvious variation; therefore it is rejected under the same rationale.
Regarding claim 19, claim 19 comprises the same and/or similar subject matter as claim 1, and is considered an obvious variation; therefore it is rejected under the same rationale.
Any inquiry concerning this communication or earlier communications from the Examiner should be directed to TODD L. BARKER whose telephone number is (571) 270 0257. The Examiner can normally be reached on Monday through Friday, 7:30am to 5:00pm.
If attempts to reach the Examiner by telephone are unsuccessful, the Examiner's supervisor Vivek Srivastava can be reached on (571) 272 7304.
/TODD L BARKER/Primary Examiner, Art Unit 2449
Prosecution Insights