Title:

System and method for providing a selectable retry strategy for frame-based communications

Document Type and Number:

United States Patent Application 20020089959

Kind Code:

Abstract:

A frame-based communications system with selectable retry strategy including a controller that programs a retry strategy field of a frame descriptor of a frame, where the retry value indicates a selected one of a normal retry count, an alternative retry count, and at least one “no retry” option. The system includes a transceiver that attempts retransmission of the frame up to as many times indicated by a selected retry count or until expiration of a frame transmit duration, or that does not attempt retransmission if the retry value indicates no retry. The no retry options may include treating the first attempt as successful or returning an unsuccessful attempt as a failure. The frame itself may be programmed with a request to not acknowledge a successfully received frame to recapture the time normally allocated to acknowledgement frames. The transceiver may include acknowledgement logic that is configured to suppress an acknowledgement frame.

Inventors:

Fischer, Michael A. (San Antonio, TX, US)
David Jr., Leach J. (San Antonio, TX, US)

Plaque It!

CROSS-REFERENCE TO RELATED APPLICATION(S)

[0001] The present application is based on U.S. Provisional Application entitled “System And Method For Synchronizing Data Transmission Across an Interface With Variable Timing”, Application No. 60/261,436 filed Jan. 11, 2001, which is hereby incorporated by reference in its entirety.

FIELD OF THE INVENTION

[0002] The present invention relates to network communications, and more particularly to a system and method for providing a selectable retry strategy for frame-based communications.

DESCRIPTION OF RELATED ART

[0003] Network communication is a growing area of technology, both for business and home applications. A network system enhances communication and provides a suitable environment for enhanced productivity and capabilities, both at home and in the workplace. It is becoming more advantageous and common for small businesses and home environments to include a local area network (LAN) that is connected to external networks, such as the Internet, that provides access to common databases and libraries and the like, and that enables communication between multiple devices that support various services, such as file sharing, printing, faxing, e-mail, voice-over-IP, video streaming, video conferencing, etc.

[0004] Many such small networks are connected through a set of wires. Wired networks are well known and generally have acceptable performance, but have many limitations, such as various cable management and convenience issues. For various reasons, wireless LAN (WLAN) technology is becoming more popular. Radio frequency (RF) appears to be the technology of choice for establishing a practical WLAN. The typical environment for wireless communications, however, is very noisy and not optimal for LAN communications. For example, most homes and work places include many electronic devices that transmit or emit RF energy resulting in an electronically noisy environment that may interfere with WLAN communications. Examples of such transmitters are microwave ovens, garage door openers and cordless telephones. Examples of unintentional emitters are radios, television sets, computer systems, etc. Further, the signal propagation characteristics of the communication medium between wireless devices constantly changes. For example, most indoor environments or rooms include multiple surfaces that are reflective to RF energy, creating multipath noise. Also, movement of items or devices or the like, such as hands, bodies, jewelry, mouse pointers, etc. or activation of electronic devices, such as cooling fans or the like, affects the overall wireless communication path and potentially degrades wireless communication performance. In summary, wireless communications must be made through a dynamic and unpredictable medium.

[0005] Wireless communications are problematic for various other reasons. The physical area served by a wireless network in not precisely defined due to the dynamic environment. In some environments, separate WLANs are proximally located which increases the likelihood for destructive interference between wireless devices that are not intended to communicate with each other. This is true because range at which WLAN radios interfere with each other is typically two to three times greater than the range at which they can reliably communicate. Power management is also an important consideration in wireless communication since wireless devices are often battery-powered. Typical solutions of increasing transmit power (or “RF power” or “radiated power”) or increasing clock speed that are often available in wired devices with ready access to utility power or the like is not usually available for wireless devices. It is not necessarily an option to decrease transmit power to reduce interference since this also reduces the communication area within a WLAN and reduces coverage faster than interference due to the square law.

[0006] Consumers are demanding high-speed wireless applications and relatively high quality of service (QoS) applications. Video applications, for example, consume four or more megabits per second (Mbps) of bandwidth. Audio applications are not as bandwidth intense, which require bandwidth on the order of 30 kilobits per second (Kbps). Nonetheless, audio applications still have many timing constraints and requirements. Audio information, for example, is very sensitive to jitter and latency variation, which if not properly addressed may result in a breakdown of communications or dissatisfied users at much lower levels at which the audio cannot be understood at all. This is particularly true for two-way communications, such as voice-over-IP and video conferencing where delay, latency and jitter issues must be addressed and resolved, which is especially difficult for wireless communications. In spite of the limited capabilities of wireless communication as compared to wired communication, consumers desire wireless devices that support these high speed timing critical applications.

[0007] The IEEE (Institute of Electrical and Electronics Engineers) 802.11-1999 standard (“the 802.11 standard”) is a protocol standard for wireless LANs. The present disclosure employs various concepts and terminology of the 802.11 standard for purposes of explanation and illustration of exemplary embodiments, although it is understood that the present invention is not limited to communications according to the 802.11 standard but instead is applicable to any communication architecture and protocol. The 802.11 standard focuses on the media access control (MAC) and physical (PHY) layer communication protocols. The basic intent of this standard is to establish communication via a wireless medium regardless of the configuration or implementation of the upper layers. In other words, the WLAN standard is an attempt to make lower communications transparent to the upper layers. Upper layer applications, however, were designed to communicate via success-oriented wired and/or optical fiber media.

[0008] Wired LANs, such as communications based on Ethernet 802.3, for example, are success-oriented and have relatively low delay and very low loss of data packets, whereas wireless communications are much less robust and have a substantially higher data loss rate. In particular, wired LANs communications typically lose less than one in one million or (1 in 10 ⁶ ) packets, whereas wireless communications based on 802.11 have a packet loss rate that is closer to one in one thousand (1 in 10 ³ ), or about three to four orders of magnitude higher loss rate as compared to wired LANs. Wired communications are much more predictable, with somewhat deterministic delays, whereas wireless communications exhibit significantly greater and less predictable delays. As used herein, the term “frame” generally denotes any type of link or physical layer data unit, and incorporates the concepts of a fixed- or variable-sized packet, cell, slot, protocol data unit (PDU), medium access control (MAC) PDU (MPDU), MAC management PDU (MMPDU), service data unit (SDU), MAC SDU (MSDU), or any other packetized means of communication.

[0009] Wired Ethernet communications use a collision detection method referred to as carrier sense multiple access with collision detection (CSMA/CD) for arbitrating access to the medium between two or more devices. Such collision detection methods are not practical in wireless communication since it is difficult for a wireless receiver to detect wireless transmission of another device while the local transmitter is operating. Wired Ethernet communications conduct retry and acknowledge functions at higher layers due to typical undetected frame loss rates of 10 ⁻⁶ . In wireless LANs, because of network media which incur frame loss rates as high as 10 ⁻³ , the retry and acknowledge functions have been incorporated into the MAC/PHY functions, and thereby consume valuable bandwidth for wireless communications. Wired communications require very little time to resolve a signal being sent. In contrast, for wireless transmissions, the receiver consumes a variable amount of valuable time to detect and resolve a signal being transmitted and to decode the information within the signal. For example, it is often necessary to measure the multipath and inter-symbol interference (ISI) distortion impact while receiving a known preamble and apply the measured distortion to the remaining packet payload in order to access the transmitted information. Valuable time may be needed to select among multiple antennas for the best signal, to set automatic gain control (AGC) levels, to synchronize the despreader, etc.

[0010] The problems with WLAN communications are compounded when implemented on personal computer (PC) platforms or the like commonly employed in home or small office environments. For example, mid- and high-layer protocol functions may be implemented using application and driver software running on a host processor, such as a central processing unit (CPU) of a PC or the like, whereas lower layer protocol functions may be implemented by firmware running on a MAC controller chip or the like mounted to an expansion board or card that is plugged into an expansion connector of the computer. This card also incorporates the physical layer (PHY) communication transceiver, such as a radio or the like, coupled between the MAC controller and one or more antennas. The variable interface between the layers above the MAC and the transceiver may include one or more input/output (I/O) buses and corresponding interface circuitry. It is imperative for proper operation that the higher layers communicate with the MAC/PHY transceiver in order to manage the information being transmitted. In the typical computer system or wireless access point (AP), a common communication mechanism between the higher layers and the transceiver is interrupt driven. Host processor interrupt latency, however, is variable, not readily determinable, and for the most part, uncontrollable by the wireless system including both the higher layer protocol software and the MAC/PHY transceiver. The timing of data transfers, interrupts, and indications between the upper-layer protocol functions and the lower-layer MAC/PHY transceiver functions, therefore, is variable and not known and subject to indeterminate delay and latency, so that the host software and drivers are unable to closely control or accurately determine the timing of the information transmission.

[0011] In the IEEE 802.11 environment, for example, the higher layer protocols handle the establishment of and bandwidth reservations for information streams with particular QoS requirements, and assumes the existence of a scheduling mechanism within the logical link or network layer, which is conceptually just above the MAC. For wireless LANs, the scheduling function is always required to achieve QoS at APs and may be required at other stations. For an IEEE 802.11 AP, this scheduler prioritizes outgoing traffic, polls other wireless stations with active QoS streams, and initiates controlled contention intervals. The scheduler delivers an appropriately ordered set of MPDUs (MAC Protocol Data Units) to the MAC transmit function for transmission during each Superframe or interval of time in conformance with the bandwidth priority, latency and other QoS criteria. A Superframe generally begins with transmission of a beacon frame followed by a contention-free period (CFP), which is then followed by a contention period (CP). The AP MAC controller performs the real time point coordination functions, transmits MPDUs, contention control (CC) frames and contention-free (CF) polls as enqueued by the scheduler, receives and validates MPDUs and reservation request (RR) frames, provides valid MPDUs to the MAC repeater of the distribution system as appropriate, controls Superframe timing using initialization parameter values provided by the scheduler or management information base (MIB), and generates acknowledgements, beacons and management frame responses in accordance with the 802.11 standard.

[0012] There are several different time bases that exist within the 802.11 AP point coordinator configuration. A first time base includes foreground tasks executed by the MAC in direct synchronism with the time base specified for intervals within 802.11 frame exchange sequences. A second time base includes background tasks activated by the MAC in response to real time events, including signals from foreground firmware, expiration of interval timers, and attention conditions when the host Input/Output (I/O) driver writes to a command register or certain other interface registers. Although background task firmware has direct access to the current 802.11 time synchronization function (TSF) timer value (a 1 microsecond (μs) time base accurate to within 4 μs at all stations in the wireless service set), processing by those tasks is subject to preemption by foreground tasks. Thus, background task processing latency varies due to WLAN traffic, host driver activity and proximity to period boundaries within the Superframe. A third time base is the host system itself, which includes an independent processor that executes the scheduler and distribution functions. The scheduler software has no control over nor ability to measure host processor interrupt response latency. This is especially problematic when the host is running a general purpose operating system, such as Windows NT or the like, rather than a real-time operating system (RTOS), because a general purpose OS is not concerned with limiting interrupt latency whereas an RTOS typically specifies an upper bound on such latency.

[0013] The scheduler is responsible for managing MPDU delivery, polling, and contention interval sequence in each Superframe, while the MAC processes outgoing frames in the order they appear on the transmit queue(s) pursuant to transmit commands from the scheduler (across the I/O interface). The MAC generates the beacon to begin the Superframe, then performs transmissions and receptions due to CF-polls and/or CC frames until the transmit queue(s) are empty or until the maximum duration for the CF interval, or CFMaxDuration, is reached. Any undelivered frames remaining on the transmit queue when the CF-End is generated at CFMaxDuration are either returned to the scheduler or discarded, depending upon the status reporting requested by the scheduler when it enqueued those frames. The scheduler generally marks frames belonging to streams with sufficiently large latency tolerance to be returned so that they may be rescheduled for transmission during a subsequent Superframe. By returning such frames to the scheduler, this rescheduling can consider the priority, latency tolerance, incurred waiting time and/or other QoS parameters defined for the stream, as well as ensuring appropriate prioritization and ordering relative to new MSDUs arriving from the distribution system, the wireless medium, or local application layer entities.

[0014] There may be a relatively short time period between the end of the CFP and the end of the Superframe. There is no guarantee that the scheduler will be able to respond fast enough to classify new arrivals, retrieve undelivered frames, make the required prioritization decisions, and load the first frame(s) for transmission during the CFP of the next Superframe between the end of a full-length CFP and the end of the Beacon that starts the next Superframe. Furthermore, after the scheduler issues the Transmit command for the first Frame Descriptor (FD) to be used during the new Superframe, several background MAC tasks have to perform some processing before that FD is ready for use by the foreground transmit task. It is noted that there is much foreground activity to preempt these background tasks, in addition to what might occur due to non-QoS traffic during the contention period near the target beacon transmission time (TBTT) when the Beacon is being prepared and transmitted.

[0015] Therefore, the scheduler needs to be able to begin submitting frames for transmission during the next Superframe prior to the end of the current Superframe and perhaps before the end of the CFP of the current Superframe. It is necessary to ensure proper allocation of frames to the intended Superframes, regardless of when the first frame for transmission during the next Superframe reaches the head of the relevant transmit queue. For example, it is necessary to achieve equivalent operation if the first frame for transmission during the next Superframe reaches the head of the relevant transmit queue before the end of the CFP of the current Superframe or after the transmission of the CF-End{+ACK} of the current Superframe but before the end of transmission of the Beacon at the beginning of the next Superframe. It is necessary to achieve proper operation when this frame does not reach the head of the transmit queue until after the end of transmission of the Beacon at the beginning of the next Superframe. Because each frame descriptor must be processed by background firmware between the time that the scheduler issues the Transmit command and that descriptor is available to the foreground MAC transmit task, the first frame of the next Superframe may reach the head of the relevant transmit queue after the end of the current Superframe even if the Transmit command for the first frame is issued before the end of the current CFP. Also, due to uncontrollable and unmeasurable (by the scheduler in real time) variations in host interrupt latency, it is not possible to ensure that the first frame of the next Superframe reaches the head of the relevant transmit queue in time even if the frame is submitted in response to a Superframe-timed interrupt, such as in response to a CF-End or a TBTT event.

[0016] Other than the sequencing problems described above, which can effect the synchronization between scheduler and MAC transmitter timing, the variable interface delay or latency hinders the scheduler's ability to perform properly various periodic functions and to monitor such periodic functions. Collocated with the scheduler is the point coordinator and the distribution services which provide AP functions. The point coordinator coordinates the flow of frames for active streams of the associated stations, which requires polling those stations for inbound frames. In particular, the point coordinator generates and enqueues polling lists and must monitor the success of the polling lists and make the necessary adjustments. In a QoS environment the scheduler is generally responsible for admittance and re-admittance of QoS frames to the set of transmit queues of the MAC at the AP and for maintaining polling lists for QoS streams. To compensate for the much greater probability of the loss of data frames on a wireless medium, the WLAN MAC protocol incurs significant overhead, including transmission of acknowledgement frames and data frame retransmissions when not acknowledged. This reduces the portion of the already limited wireless bandwidth that is available for user data transfers.

[0017] It is desired to implement wireless communications for all types of protocols and architectures that is capable of meeting arbitrary bandwidth and QoS demands by consumers. It is desired to implement wireless communication devices that efficiently utilize the wireless medium to establish and maintain successful wireless communications for many applications, including high bandwidth and latency sensitive voice, video, and multi-media applications. It is desirable to provide service and priority differentiation so that the QoS scheduler function may enforce a bandwidth allocation policy specified by the network operator. Such differentiation, for example, would enable allocation of bandwidth to subscribers who pay for a “premium service” in preference to those who subscribe to a “basic service”.

SUMMARY OF THE INVENTION

[0018] A method of providing a selectable retry strategy for frame-based communications according to an embodiment of the present invention includes programming a retry value to indicate a selected one of no retry or a retry count, transmitting a frame that is associated with the selected retry value, and suppressing retransmission of the frame regardless of the retry count if the retry value indicates no retry.

[0019] The normal operation of many communications systems is to retry transmission of a frame at least once if that frame is not successfully received. A standard or normal retry count may be specified to limit the number of retries, where the normal retry count is used by default for each frame. For example, a management information base (MIB) for wireless communications according to the 802.11 standard may specify a normal retry count. An acknowledgement frame or the like may be sent by the receiving device to the transmitting device for each frame to acknowledge successful reception of the frame. In some applications, however, such as those that use a real-time video stream or the like, there may be insufficient time to retry the frame. It may be too late, for example for the receiver that is displaying the video stream to use the failed frame by the time it is received. The selectable retry strategy, by selection of no retry, is useful in such applications in which it would otherwise be wasteful to consume time on the medium retrying a frame that will not be used even if it were to be received successfully.

[0020] The method may further include programming the retry value to indicate a selected one of no retry, a first retry count or a second retry count. The first retry count, for example, may be a normal retry count that would otherwise be applicable to all frames, such as a normal or standard retry count specified in an 802.11 MIB. The use of an alternative retry count provides at least one benefit in that a controller or host software may program a different retry count for specific frames depending on the particular application. A significantly smaller number of retries may be used, for example, for latency sensitive frames. Of course, the alternative retry count may be programmed to any reasonable number. The no retry indication may mean that a first attempt is treated as successful so that retransmission is not attempted.

[0021] The retry value may indicate treating a first attempt as successful and not retrying transmission or returning an unsuccessful transmission attempt as a failure and not retrying transmission. In the first case, the transmission is treated as successful since there may be no benefit in retransmitting the frame even if unsuccessfully received. In the second case, the transmitting device determines if an acknowledgement frame is received, and if not, reports the failure back to a higher level, such as a network driver or the like. In either case, transmission is only attempted once and not retried.

[0022] The method may further include specifying a frame lifetime for the associated frame that indicates a maximum retry time and attempting re-transmission until expiration of the frame lifetime or as many times specified by the retry count, whichever occurs first. The retry count may be a normal retry count or an alternative retry count.

[0023] In one embodiment, the retry strategy is implemented using two bits to achieve up to four optional indications, including a normal retry count, an alternative retry count, treating a first attempt as successful and not retrying transmission, or returning an unsuccessful transmission attempt as a failure and not retrying transmission. These above-listed four options are exemplary only and additional indications are contemplated that may be used instead or in addition in which additional retry strategy bits may be used. The method may include programming a retry strategy field of a frame descriptor for each frame to be transmitted to indicate the selected retry strategy. A frame descriptor is provided for each frame to include control and status information for the frame. Thus, a selectable retry strategy may be programmed for each frame.

[0024] The method may further include programming the retry value to indicate treating a first attempt as successful and not retrying transmission and programming the frame to indicate that acknowledgement (ACK) is not requested. This allows an additional benefit in that the acknowledgement response may also be selectively suppressed. The method may further include transmitting a second frame before completion of an acknowledgement period of time. In this case, the next outgoing frame can be transmitted without waiting for the acknowledgement frame thereby eliminating the time for the acknowledgement frame and its inter-frame gap. It is noted that programming the frame with a selectable acknowledgement request may not be supported by some devices, such as those according to the 802.11 standard in which the protocol was standardized without provision for a selectable ACK function. Thus, may be desired to encode the selectable acknowledgement request in the frame in such a manner that it will be ignored by those devices that do not support the selectable ACK option. In a particular embodiment, the method includes programming at least one bit in a duration/ID field of the frame. The low order bits of the duration/ID field of the MAC header are particularly advantageous in that those devices or stations that do not support the selectable ACK option will ignore the programmed bits (if the high order two bits of this field are each “1”) and thus not be affected. A separate field may be used to incorporate the retry strategy, such as a separate Quality of Service (QoS) field or the like.

[0025] The method described herein is applicable to any type of frame-based communication system, including those that use wired or wireless mediums. It is noted that the present invention is particularly advantageous to wireless LAN communications, which are typically very noisy and not optimal for LAN communications. A selectable retry strategy allows significant improvement for time-sensitive or latency-sensitive applications in which the standard or normal retry approach is insufficient or inefficient, such as, for example, when the time required to utilize the normal approach may cause unacceptable delay to other frames destined for other stations that are enqueued behind the frame being retried.

[0026] A method of improving usage of a wireless medium according to an embodiment of the present invention includes applying, by a first transceiver system, a no retry strategy to a first frame so that retransmission of the first frame will not be attempted, programming the first frame with an indication that acknowledgement is not requested, and transmitting the first frame via the wireless medium. The method further includes successfully receiving, by a second transceiver system, the first frame via the wireless medium, detecting the no acknowledgement indication, and suppressing sending an acknowledgement frame in response to successfully receiving the first frame. The usage of the wireless medium is improved by eliminating frame retries and further eliminating acknowledgement frames in response to successful reception. The method may further include programming at least one bit of a duration/ID field of the first frame with the acknowledgement request. Alternatively, the method may further include programming at least one bit of a Quality of Service control field of the first frame with the acknowledgement request.

[0027] The method may further include transmitting, by the first transceiver system via the wireless medium, a second frame prior to expiration of a predetermined inter-frame gap period, or prior to expiration of a predetermined acknowledgement period after transmission of the first frame. In this manner, data throughput of the wireless medium may be increased for selected applications since the time reserved for the acknowledgement frame and its interframe gap may instead be used for frame transmission.

[0028] A frame-based communications system with selectable retry strategy according to an embodiment of the present invention includes a controller that programs a retry value associated with a frame, where the retry value indicates a selected one of no retry and a retry count, and a transceiver that transmits the frame at least once, that attempts retransmission of the frame up to as many times indicated by the retry count if the retry value indicates the retry count, and that does not attempt retransmission of the frame if the retry value indicates no retry. The controller may include a scheduling entity that programs a retry value for each frame to be transmitted. The transceiver may include a queue and a frame manager that receives and enqueues frames into the queue and that determines the retry value for each enqueued frame. The scheduling entity may be configured to program a retry strategy field of a frame descriptor for each frame to be transmitted. The frame manager may be configured to determine the retry value for each frame from the retry strategy field of a corresponding frame descriptor. The controller may be configured to program the retry value to indicate a selected retry strategy in a similar manner as previously described.

[0029] The controller may further be configured to program a frame lifetime that specifies a maximum time for attempting retries for the frame. In this manner, the transceiver is configured to transmit the frame at least once and to attempt retransmission of the frame until expiration of the frame lifetime or for as many times indicated by a retry count, if specified, whichever occurs first.

[0030] The transceiver may further include a transmission scheduler that programs a frame for transmission with an acknowledgement request indicating whether acknowledgement of successful receipt of the frame is requested. The duration/ID field of the frame or a separate QoS control field may be used for programming the selectable acknowledgement request. The transmission scheduler may be configured to program the frame with a no acknowledgment request and to schedule transmission of a subsequent frame before expiration of a predetermined acknowledgement period. In this case, it is assumed that the receiving device is configured to detect the no acknowledgement request and not send an acknowledgement frame. In one embodiment, the transceiver may include receive logic that is configured to send an acknowledgement frame by default in response to a successfully received frame, and acknowledgement logic that is configured to instruct the receive logic to suppress sending an acknowledgement frame if an acknowledgement request of a successfully received frame indicates that acknowledgement is not requested.

[0031] Many different embodiments or options are contemplated without departing from the spirit and scope of the present invention. The controller and transceiver may be implemented as a wireless access point (AP). The controller may be implemented on a computer system, which further includes a memory that stores application software, a processor that executes the application software to perform controller functions, and an expansion bus system. In this case, the transceiver may be implemented on an expansion card that interfaces the expansion bus system of the computer system. The transceiver may include a host interface, a media access control (MAC) device, and a radio that enables wireless communications via a wireless medium. The MAC device may include a queue, a frame manager that receives and enqueues frames into the queue and that detects a retry value for each frame, and a transmission scheduler that dequeues frames from the queue and that transmits frames in accordance with an associated retry value. The transmission scheduler may be configured to return an unsuccessful indication, such as to the host, if an acknowledgement frame is not received within a predetermined period of time for a frame transmitted once and having a no retry and return unsuccessful attempt as failure indication.

[0032] The transmission scheduler may include retry logic that selectively programs frames for transmission with an acknowledgement request. For those frames having a retry value of no retry and treat first attempt as successful indication, the retry logic may selectively program an acknowledgement request for that frame to indicate that acknowledgment is not requested. The MAC device may further include acknowledgement logic that is configured to instruct the receive logic to suppress sending an acknowledgement frame if an acknowledgement request of a successfully received frame indicates that acknowledgement is not requested.

BRIEF DESCRIPTION OF THE DRAWINGS

[0033] A better understanding of the present invention can be obtained when the following detailed description of the preferred embodiment is considered in conjunction with the following drawings, in which:

[0034] FIG. 1 is a simplified block diagram of an access point (AP) within a wireless communication system implemented in accordance with an embodiment of the present invention.

[0035] FIG. 2 is a block diagram of a computer system configured as an exemplary embodiment of the AP of FIG. 1 .

[0036] FIG. 3 is a more detailed block diagram of the WLAN card interfaced to the host system of FIG. 2 .

[0037] FIG. 4 is a simplified diagram of an exemplary frame and frame descriptor.

[0038] FIGS. 5 A- 5 C are simplified block diagrams of the transmission logic of the MAC device of FIG. 2 illustrating persistent frame operation.

[0039] FIG. 6 is a simplified block diagram illustrating operation between the host driver and the MAC device of FIG. 2 for clearing a persistent frame.

[0040] FIGS. 7 A- 7 C show an individual transmit queue of FIG. 2 operating in a manner that illustrates the benefits of persistent frame capabilities for submitting polling lists employing polling frames marked as persistent.

[0041] FIGS. 8A and 8B are simplified block diagrams of the transmission logic of the MAC device of FIG. 2 illustrating exemplary queue mark (QM) operation employing the QM field of frame descriptors and optional QM bits.

[0042] FIGS. 9A and 9B are simplified block diagrams of the transmission logic of the MAC device of FIG. 2 illustrating an alternative embodiment of the QM operation.

[0043] FIG. 10 is a partial block and timing diagram of the transmission logic of the MAC device of FIG. 2 illustrating control capability employing QM operation while there is sufficient time in a given interval I 1 .

[0044] FIG. 11 is a partial block and timing diagram of the transmission logic of the MAC device of FIG. 2 illustrating QM operation when the MAC device does not have time to transmit all frames intended to be sent during the interval I 1 .

[0045] FIGS. 12A and 12B are partial block and timing diagrams of the transmission logic of the MAC device of FIG. 2 illustrating QM operation when the host driver of FIG. 2 is too slow and does not submit frames into the transmit queue of FIG. 3 in time for transmission during the current interval I 1 .

[0046] FIG. 13 is a tabular diagram illustrating the retry strategy programmed within the RS field of a frame descriptor of a frame.

[0047] FIG. 14 is a simplified block diagram of a transceiver that is configured to detect a selectable acknowledgement request in successfully received frames.

[0048] FIG. 15 is a flowchart diagram of an exemplary routine of the transmission scheduler of the MAC device of FIG. 2 for processing frames within any of the transmit queues of FIG. 3 .

[0049] FIG. 16 is a SDL process diagram that describes the behavior of QM processing within the MAC transmission logic.

DETAILED DESCRIPTION OF EMBODIMENT(S) OF THE INVENTION

[0050] FIG. 1 is a simplified block diagram of an access point (AP) 100 within a wireless communication system. The AP 100 includes a station host or AP controller 101 and a wireless network transceiver 103 that communicate in a wireless medium 106 via at least one antenna 104 . It is noted that the AP 100 is also representative of the applicable functionality of a wireless station in accordance with embodiments of the present invention. In the case of a station, the AP controller 101 is typically a personal computer (PC), wireless information appliance, or the like, with various subsystem functions performed by software executing on a processor that is also used to perform other functions of the station. In the case of an AP, the AP controller 101 is typically a dedicated processor that only performs the network-related functions, although there are embodiments of an access point in software that runs on a PC. The more extensive set of functions for illustrating the present invention are utilized at an AP, so hereafter the references are to an AP, with the understanding that the equivalent functions, or a subset thereof, may also exist at a station. In the AP embodiment, the AP 100 communicates with a distribution system 102 via an interface 108 .

[0051] The AP controller 101 and the transceiver 103 communicate across an internal interface 105 , which introduces indeterminate and generally uncontrollable delay of information being transferred, and is thus referred to as a “variable delay” interface. In particular, the AP controller 101 submits fixed- or variable-sized data units, cells, packets or frames, generally referred to as “frames”, via the variable delay interface 105 to the transceiver 103 for transmission. The AP controller 101 may also send command frames or the like to the transceiver 103 . In accordance with embodiments of the invention, as further described below, the AP controller 101 further submits frame descriptors that define various transmission policies to be performed by transmitter functions of the transceiver 103 . The transceiver 103 receives the frames and frame descriptors from the variable delay interface 105 and transmits the frames onto the wireless medium 106 in accordance with programmed parameters within the frame descriptors. The transceiver 103 also receives frames of information from the wireless medium 106 via the antenna 104 and provides the received frames to the AP controller 101 across the variable delay interface 105 . The transceiver 103 may also report status information to the AP controller 101 across the variable delay interface 105 . The status information may include for example an indication of whether frames have been transmitted successfully or not.

[0052] The particular configuration and implementation of the AP controller 101 depends upon the type of communication network, its data transfer bandwidth and the type and amount of information being processed. In the AP embodiment, the AP controller 101 is a management and frame forwarding entity and coordinates functions across the wireless medium 106 with other network-attached devices known as stations. For example, the AP controller 101 may include both station functionality and provide access to distribution services on behalf of stations communicating via the wireless medium 106 . A common instance is the AP 100 and associated stations according to the IEEE 802.11 standard for wireless LANs. In an exemplary 802.11 configuration, the AP controller 101 may further perform point coordination functions (PCF), which are a class of possible coordination functions in which the coordination function logic is active in only one station (required to be the AP in 802.11) in the basic service set (BSS) at any given time that the network is in operation. References to the 802.11 standard and associated operation is exemplary only, where it is understood that the present invention is not limited to 802.11 and may apply to any appropriate wireless communication protocol.

[0053] In the embodiment shown, the AP controller 101 includes a bandwidth manager 107 and a scheduling entity 109 . The transceiver 103 includes a medium access control (MAC) function 111 , which further includes transmission control logic 113 for sending frames and reception control logic 112 for receiving frames. The term “logic” collectively refers to any combination of circuitry and programs, such as software and firmware and the like configured to perform a related set of one or more functions. The reception control logic 112 and the transmission control logic 113 are coupled to a physical-layer (PHY) device 115 of the transceiver 103 for performing wireless communications via the antenna 104 . The AP controller 101 ultimately manages the communications conducted by the transceiver 103 across the variable delay interface 105 . In many system configurations, the transfer timing across the variable delay interface 105 is not tightly controlled, resulting in substantial and unknown transfer delay that significantly mitigate the ability of the scheduling entity 109 to perform accurate and efficient management of wireless communications by the transceiver 103 . The variable timing may be due to interference of hardware or system software or both.

[0054] In many network configurations, the distribution system 102 and higher layer communication protocols are not aware that a link of the communications is conducted wirelessly. In an 802.11 configuration, for example, the distribution system 102 and its higher layer protocols are success-oriented as is typical with wired networks, whereas the wireless medium 106 exhibits substantially increased latency and frame loss rate as compared to wireless media. The dynamic and unknown latency across the variable delay interface 105 prevents the AP controller 101 from tightly controlling operations of the transceiver 103 . This in turn prevents the AP controller 101 from effectively managing time-critical aspects of the communication conducted via the transceiver 103 across a wireless medium. Without a mechanism to compensate for the effect of the variable delay interface 105 , the efficiency and perhaps the interoperability of the wireless network can be impaired.

[0055] Embodiments of the present invention are directed towards aspects of the transmission control logic 113 that are directly controlled by the scheduling entity 109 to coordinate and improve communication between the scheduling entity 109 and the transmission control logic 113 across the variable delay interface 105 . In one common application, the bandwidth manager 107 and the scheduling entity 109 cooperate to establish and manage quality of service (QoS) admission control, congestion control, prioritization and the like to establish and enforce bandwidth reservations for various information streams and services utilizing the network. The AP controller 101 operates on a substantially different time base as compared to the transceiver 103 . Furthermore, higher layer protocols used to manage the distribution system 102 operate in an arbitrary, distributed time base, such as on the order of several milliseconds (ms) or the like. The distribution system 102 is generally asynchronous in nature and operates on global and human-based timeframes and generally manages overall bandwidth allocations and QoS contracts to ensure that information, such as audio and/or video streams of information, are delivered within particular predetermined time constraints, independent of network load of “best effort” data traffic. In general, the distribution system 102 is network agnostic and attaches to a network independent end system that communicates with other end systems that communicate with each other regardless of the particular network configuration through which they are coupled. The distribution system 102 also incorporates intermediate network systems that are stream and service specific.

[0056] In contrast, the wireless transceiver 103 operates with much more specific timing constraints on the order of several microseconds (μs) or less. The transceiver 103 must be relatively accurate and must maintain synchronism with wireless network timing within tight timing constraints in order to establish and maintain communication with other wireless devices. For the 802.11 standard, station transceiver synchronism must be maintained to +/−2 μs. Failure to maintain communication protocols and timing constraints at the MAC level results in failure of communication. The wireless medium 106 , however, is dynamic and unpredictable. The transceiver 103 must use a wireless communication protocol that includes substantial overhead to overcome characteristics of the wireless medium 106 . Furthermore, the transceiver 103 must perform substantial processing in order to measure and quantify the status of the wireless medium 106 , such as measuring multipath and other distortion, to determine the distortion in order to accurately decode or demodulate transmitted frames. For example, each frame typically has a known preamble so that the receiver may measure the distortion effects on the preamble and apply the measured distortion to the remainder of the transmitted frame.

[0057] In accordance with embodiments of the present invention, many of the communication functions traditionally conducted solely or partially by the scheduling entity 109 are effectively transferred to the transmission control logic 113 . In this manner, the AP controller 101 is able to maintain more accurate control and to perform more efficient management of scheduling, coordination and QoS functions that would otherwise not be possible due to the variable delay interface 105 . The transmission control logic 113 includes one or more dynamic functions that are under direct control by the scheduling entity 109 . In illustrated embodiments, for example, the scheduling entity 109 submits a frame descriptor (FD) with each frame, where the frame descriptor includes one or more programmable fields that instruct the transmission control logic 113 how to handle the corresponding frame. The frame descriptor may be prepended to the frame and transferred to the transmission control logic 113 via the variable delay interface 105 . The frame descriptors are not transmitted with the frames, but instead are employed to instruct the transmission control logic 113 regarding transmission of the frames and the reporting of status about the frames.

[0058] FIG. 2 is a block diagram of a computer system 200 configured to provide AP functionality for purposes of illustrating exemplary embodiments of the present invention, and is a PC specific embodiment of the AP 100 . The computer system 200 may be any type of computer system, such as a desktop computer, a portable computer, a laptop computer, or any type of smaller or portable type of computing device, such as a personal digital assistant (PDA) or the like, or any type of embedded computer or processor as known to those skilled in the art. The computer system 200 includes a central processing unit (CPU) 201 , which is a general purpose digital processor, zero or more storage devices 205 , and a memory system 203 coupled to bus and support system 207 . The memory system 203 may include any combination of memory devices, such as dynamic random access memory (DRAM), static RAM (SRAM) devices, programmable and non-programmable read only memory (ROM) devices, etc. The storage 205 may include any type of read or read/write (RIW) data storage devices such as floppy drives, disk drives, tape drives, etc. The bus and support system 207 includes any combination of one or more bus and interface circuits and system support logic appropriate for the particular type of computer system 200 . For desktop systems, the bus and support system 207 may include one or more peripheral component interconnect (PCI) buses, one or more industry standard architecture (ISA) buses, universal serial buses (USBs), etc., with one or more corresponding expansion connectors or slots 209 as known to those skilled in the art. For portable computer systems and smaller for factors, the expansion connectors 209 are often implemented as PCMCIA, PC Card slots, compact Flash slots or the like.

[0059] To perform the AP function, the computer system 200 includes a local area network (LAN) card 211 for attaching the computer system 200 to a wired LAN 213 which serves as the distribution system 102 . For any type of computer system 200 (station or AP), a wireless LAN (WLAN) card 215 is attached into an appropriate expansion connector 209 for interfacing to the computer system 200 to include wireless communication capabilities. The WLAN card 215 includes a host interface (IF) 221 that couples to the bus and support system 207 via the expansion connector 209 . The host interface 221 is coupled to a MAC device 223 for performing the MAC function 111 , which is further coupled to a radio 225 for performing the PHY device 115 functions. The radio 225 includes at least one antenna 227 for communications on the wireless medium 106 , similar to the antenna 104 .

[0060] The MAC device 223 includes a transmit (TX) control and scheduler system 231 including one or more transmit queues for receiving outgoing FDs and frames from the host interface 221 and enabling transmission of the frames via the radio 225 and the antenna 227 . The TX control and scheduler system 231 performs the functions of the TX control logic 113 . Frames received from other wireless devices via the wireless medium 106 by the radio 225 via the antenna 227 are handled by a receive (RX) system 235 for validation, address recognition, and, if addressed to this station or AP, for delivery to the computer system 200 via the host interface 221 . A portion of the memory system 203 is typically loaded with an operating system (O/S) 217 , which further mediates communication between application programs or software 218 and a network I/O driver 219 for communicating with the WLAN card 215 . The operating system 217 may comprise, for example, various Windows configurations by MicroSoft, such as Windows 95, 98, ME, 2000, NT, etc. Other suitable operating systems are contemplated, such as Novell Netware or the like. The operating system 217 further loads and manages one or more application software or programs 218 for conducting wireless communications utilizing the WLAN card 215 via the network I/O driver 219 and including AP software to perform the functions of the bandwidth manager 107 and the scheduling entity 109 .

[0061] The application programs 218 , the operating system 217 , and the network V 0 driver 219 , together form an exemplary embodiment of the AP controller 101 previously described, including appropriate AP software. It is understood, therefore, that references to the scheduling entity 109 and the bandwidth manager 107 in association with the computer system 200 are indicative of the CPU 201 executing the operating system 217 , the application programs 218 and the network I/O driver 219 from the memory system 203 performing the relevant functions. The operating system 217 , the network I/O driver 219 , the bus and support system 207 and the host interface 221 generally form an exemplary embodiment of the variable delay interface 105 . Thus, application programs 218 must communicate through the variable delay interface 105 in order to manage wireless communications conducted by the WLAN card 215 and controlled by the MAC device 223 . Yet the operating system 217 is often not a real-time operating system (RTOS) and is therefore unable to provide for tight and predictable timing of communications between the application programs 218 and the expansion connectors 209 , particularly in Windows-based systems. Even if the operation system 217 is an RTOS, the granularity is typically not sufficient for that needed by the wireless MAC protocol. Furthermore, the delays through the bus and support system 207 , expansion connectors 209 and host interface 221 are often variable. As a result, the TX control and scheduler 231 is implemented with additional programmable capabilities to enable and improve communications between the application programs 218 and the MAC device 223 .

[0062] FIG. 3 is a more detailed block diagram of the WLAN card 215 as interfaced to the network I/O driver 219 via the host I/O system 207 , 209 . The MAC device 223 interfaces the host interface 221 , which transfers frames and frame descriptors (FDs) for transmission from the host driver 219 to an input queue (IN Q) 301 . A transmit (TX) frame manager 303 retrieves frames and FDs from the IN Q 301 and enqueues each frame and FD into one of several transmit queues 305 , individually labeled Q 0 , Q 1 , Q 2 , Q 3 . . . QN, where “N” is any positive integer, although a single transmit queue 305 is contemplated as well. Any one or more of the transmit queues 305 may be configured or operated as first-in first-out (FIFO) queues, although other types of queues are contemplated. Regardless of whether a transmit queue is configured or operated as a FIFO queue, any one or more of the queues may allow non-FIFO removal behavior based on other properties of queued elements, such as priority, destination address, frame type, etc. Also, a persistent queue, labeled QP, is contemplated in which all frames enqueued therein are considered persistent frames. In one embodiment, a separate persistent queue QP is provided so that each frame enqueued thereon, as instructed by the controller or scheduler, is considered a persistent frame until deleted from the queue QP. Alternatively, any of the transmit queues 305 may be temporarily or permanently programmed as a persistent queue QP.

[0063] In the embodiment shown, the transmit queues 305 are organized according to level of priority. In particular, the first queue Q 0 is used to hold the lowest priority pending transmissions intended for best effort frames. A next priority queue Q 2 is intended for medium priority frames. Higher numbered queues, such as Q 3 -QN are intended for high-priority traffic. In a particular 802.11 embodiment for example, the low priority queue Q 0 is intended for best effort MPDU and for MMPDUs and the like to be transmitted during the contention period (CP). Q 2 is for transmission of frames of high priority during the contention free period (CFP) and intended for contention-free asynchronous delivery and for CF-poll frames of stations or the contention free (CF) polling list. The high-priority queues beginning with Q 3 are for frames to be transmitted first during CFP and intended for latency-sensitive or jitter-sensitive traffic. The TX frame manager 303 detects the type and priority of each frame pulled from the IN Q 301 based on information in the FD, and inserts the frame at the end of an appropriate one of the transmit queues 305 .

[0064] Each transmit queue 305 has sufficient capacity for storing multiple frames or MPDUs in first-in, first-out order for transmission. Each transmit queue 305 may further include a provision for storing a frame descriptor for each frame, where the frame descriptor, further described below, includes various parameters as to how or when the corresponding frame is to be transmitted. Additionally, each transmit queue 305 may provide storage of a programmable marker for each frame, referred to as a queue mark (QM) bit or the like, that is used for QM operation as described further below. QM operation allows a correspondence to be established between a point in the frame sequence from the scheduling entity 109 running on its time base and on its side of the variable delay interface 105 with a point in the MAC protocol sequence, such as the start of a particular interval, in the time base of the MAC controller and on the opposite side of the variable delay interface 105 . This function of QM is reflected in its context-dependent usage, which is sometimes delay, sometimes discard, sometimes start, sometimes stop, etc. Additionally, each transmit queue 305 may provide storage of a programmable persistence flag or marker for each frame, such as a persistence bit or the like, that is used to implement frame persistance as described further below.

[0065] The outputs of the transmit queues 305 are provided to a transmit scheduler 307 , which schedules frames from the transmit queues 305 for transmission via a transmit function (TF) 309 . The transmit function 309 provides frames for transmission via the modem interface 311 , which conveys the frames to the radio 225 for transmission via the antenna 227 . Frames received by the radio 225 are provided to a receive function (RF) 313 via the modem interface 311 , and are then provided to receive logic 315 . The receive logic 315 provides the received frames to the network I/O driver 219 through the host interface 221 . Additional detail of the receive (RX) logic 315 will not be further described herein other than reference to acknowledge (ACK) logic 316 described further below. Access and response logic 317 is coupled to the modem interface 311 , the transmit function 309 , the receive function 313 and the transmission scheduler 307 for controlling wireless communications and facilitating coordination between transmitting and receiving frames. The transmission scheduler 307 conveys frames via a transmit (TX) done queue 319 to the host interface 221 for purposes of completion status reporting to the host, in such a manner that decouples the rate at which the host queries such status from the rate at which the MAC device 223 processes FDs. For example, frames that are retrieved from the transmit queues 305 and that bypass the transmit function 309 (e.g., not transmitted or “dropped”), are placed by the transmission scheduler 307 onto the TX done queue 319 with a status bit set to indicate that the frame was not delivered. As described further below, the transmission scheduler 307 includes a re-enqueue path 321 to the TX frame manager 303 for re-enqueuing persistent frames as further described below. It is noted that transmission and re-enqueuing operations may be considered independent operations and it is not intended that they be performed in any particular order.

[0066] As described previously, a common communication mechanism between the network I/O driver 219 and the MAC device 223 is based on interrupts. Host interrupt latency is variable, unknown and for the most part, uncontrollable by the software. Therefore, the timing of communications, frame transfer and indications between the application programs 218 and the MAC device 223 is variable and not known. Improved communications across the variable delay interface 105 , such as including the operating system 217 , the network I/O driver 219 , the bus and support system 207 and the expansion connectors 209 , are described herein so that the applications and network I/O driver 219 are able to manage properly information transmission by the MAC device 223 . As further described below, each frame descriptor associated with a corresponding frame forwarded by the network I/O driver 219 to the MAC device 223 includes a queue mark (QM) field that serves as a timing index that allows the transmission scheduler 307 to determine whether to transmit (or to drop) one or more frames for purposes of synchronizing the sequence of frames to the transfer intervals defined by the MAC protocol. This timing index allows the transmission scheduler 307 to realign timing to what is intended by the scheduling entity 109 on the host side of the variable delay interface 105 .

[0067] The frame descriptor further includes a persistence (PRST) field that instructs the transmission scheduler 307 to re-enqueue the corresponding frame after processing by submitting the frame back to the TX frame manager 303 via the re-enqueue path 321 . The frame descriptor includes a retry strategy (RS) field that instructs the TX frame manager 303 regarding the retry strategy for the corresponding frame, such as whether to retry the frame in the event that the initial delivery attempt is unsuccessful, and if unsuccessful, how many times to retry the frame. The frame descriptor may further include a frame lifetime (FL) field that includes a timing parameter that specifies a retry time duration. The retry time duration may be used instead of a retry number or in addition thereto. If a retry count is specified along with a frame lifetime, the frame is retried up to the specified number of times defined by the retry count or until expiration of the frame lifetime, whichever occurs first.

[0068] The transmission scheduler 307 optionally includes retry logic 308 that modifies a frame based on the programmed value within the RS field of the frame descriptor of the frame. In one embodiment, the retry logic 308 programs the duration/ID field of the MAC header information of the frame to be transmitted with at least one acknowledgement request (AR) bit that indicates to the receiving device whether to transmit an ACK frame to indicate successful reception. The duration/ID field may include the same bits of the RS field to indicate the retry strategy and the acknowledgement request. Alternatively, a separate field may be employed, such as a Quality of Service (QoS) control field or the like, to specify the retry strategy. The retry strategy function is relevant if the MAC protocol includes MAC-layer acknowledgement and retry of unacknowledged frames. These are not traditionally used at the MAC layer, but are often added in MAC protocols intended for use over wireless physical layers, such as 802.11. A retry strategy control for number retry attempts is useful with any MAC protocol that includes ACK and retries. An additional benefit, however, is achievable if it is possible within the protocol to selectively suppress the generation of the ACK response for the transmissions that are not going to be retried even if the transmission is unsuccessful. An example of this are frames of a real-time video stream for which there is typically insufficient time to perform the retry, so the visible effect to the viewer is reduced if the subsequent frames are not delayed by retries of previous frames. These cases allow further improvement of throughput because if the sending station knows that no ACK will be returned, then the next outgoing frame can be transmitted without waiting for the ACK frame, and eliminating the time normally reserved for the ACK frame and its corresponding inter-frame gap. The 802.11 protocol was standardized without provision for a selective ACK function. In one embodiment, a “do not ACK” message is encoded into an existing field of frame to be transmitted where the message is ignored by stations that do not support the option. A good place to do this for 802.11 contention free transmissions are bits within the low order 14 bits of the duration/ID field of the MAC header when the highest order 2 bits are both set to logic “1”. During normal operation, a receiving device transmits an ACK frame a short inter-frame space (SIPS) period after successfully receiving a frame from a transmitting device. As further described below, the ACK logic 316 in the RX logic 315 examines the received frame and does not transmit an ACK frame if the frame was valid and addressed to this station, but the acknowledgement request bit indicates that an ACK frame is not requested.

[0069] FIG. 4 is a simplified diagram of an exemplary frame 401 and a frame descriptor 403 implemented according to an embodiment of the present invention. In the embodiment shown, the frame descriptor 403 is appended to (or otherwise associated with) the frame 401 by the network I/O driver 219 or other higher layer logic or software and transferred to the MAC device 223 via the variable delay interface 105 . The frame descriptor 403 further includes a TX control field 405 , which further includes one or more fields programmable by the network I/O driver 219 and/or application programs 218 for purposes of controlling transmission via the MAC device 223 .

[0070] As shown, the TX control field 405 includes a retry strategy (RS) field for defining one of several selectable retry strategies for the frame 401 . The TX control field 405 further includes a frame lifetime (FL) field that includes a retry timing value that specifies a maximum amount of time to retry a frame if it is to be retried. The controller or scheduler may program the FL field to be used alone or in combination with the retry strategy. For example, a retry duration may be used instead or to override any specified retry count so that the associated frame is retried until expiration of the specified frame lifetime. Or, the retransmission of the frame may be attempted until expiration of the frame lifetime or as many times as indicated by the retry count, whichever occurs first. A null value may be programmed into the frame lifetime field if a lifetime is not to be used.

[0071] The TX control field 405 includes a persistence (PRST) field for marking the frame 401 as a persistent frame. The TX control field 405 further includes a queue mark (QM) field that marks the frame 401 as a QM frame for purposes of establishing a reference point in the queued sequence of frames that is to be transmitted in synchronism with the MAC timing for a particular interval, or not transmitted if such synchronism cannot be established. The QM field may comprise a single bit to mark the corresponding frame for QM operation. It is noted that when a frame is said to be marked for QM operation, making that frame a QM frame, it is understood that the QM field of the corresponding frame descriptor contains a bit or code value that indicates the QM operation. In one embodiment, the QM field indicates that the frame 401 is to be transmitted as the first frame in the next instance of a particular type of transmission interval. In another embodiment, the QM field denotes the frame as the last transmitted frame in a current interval. A QM frame may or may not be intended for transmission. The present invention contemplates any of these variations for programming the QM field.

[0072] FIGS. 5 A- 5 C are simplified block diagrams of a portion of the MAC device 223 illustrating persistent frame operation. As shown in FIG. 5A, a selected one of the transmit queues 305 is loaded by the TX frame manager 303 with six frames supplied by the network I/O driver 219 , F 1 , F 2 , F 3 , F 4 , F 5 and F 6 in order so that F 1 is intended to be transmitted first and F 6 last. Another frame F 7 is being provided by the network I/O driver 219 , such as the next frame in the IN Q 301 . In one embodiment, the transmit queue 305 is a FIFO queue, so that the intended transmission order is from right to left where the transmit queue 305 effectively operates as a linear buffer. The transmit scheduler 307 dequeues each frame one at a time for submission to the transmit function 309 . The transmit scheduler 307 also dequeues and inspects each corresponding frame descriptor. Thus, when the transmit scheduler 307 is operating from the transmit queue 305 as shown, it dequeues the frames F 1 , F 2 , F 3 , F 4 , F 5 and F 6 in that order for delivery to the transmit function 309 for transmission.

[0073] The transmit scheduler 307 includes persistence logic (PL) 501 that detects persistent frames to enable persistent frame operation. As shown, persistent frames, such as a first frame F 1 , are indicated by a persistent indicator “P”, where persistent frames are indicated in any one of several ways. In one embodiment, the frame descriptor of a frame has its PRST field programmed as a persistent frame. In another embodiment, a persistent frame bit or the like programmed into the corresponding transmit queue 305 by the TX frame manager 303 when the frame is enqueued. In another embodiment, the frame is considered persistent when enqueued into a persistent queue, such as the queue QP or any transmit queue 305 that is programmed as persistent. In another embodiment, any frame of a certain frame type may automatically persistent frames, such as polling frames or the like. The persistence logic 501 is configured to detect persistent frames according to any one or more or a combination of all of these methods depending upon the particular configuration desired.

[0074] As shown in FIG. 5 B, the transmit scheduler 307 dequeues the next frame F 1 from the transmit queue 305 and the persistence logic 501 detects that the frame F 1 is persistent and asserts a persistent signal indicative thereof. It is noted that the persistence logic 501 may be implemented in any of many ways, such as by firmware or by logic either separate from or incorporated within the transmit scheduler 307 . The transmit scheduler 307 detects the persistent signal and identifies the frame as persistent. The transmit scheduler 307 submits the frame F 1 to the transmit function 309 for transmission as shown at 505 , excluding the frame descriptor. Furthermore, the transmission scheduler 307 copies the persistent frame F 1 and its frame descriptor back to the TX frame manager 303 via the re-enqueue path 321 as shown at 507 . In this case, the next frame F 7 supplied by the network I/O driver 219 was added to the end of the transmit queue 305 by the TX frame manager 303 by the time the re-enqueuing of frame F 1 occurred.

[0075] As shown in FIG. 5 C, the TX frame manager 303 re-enqueues the persistent frame F 1 into the transmit frame 305 as shown at 509 . The corresponding frame descriptor is also enqueued so that the frame maintains its persistent status. Alternatively, if a persistent bit is included in the transmit queue 305 , the TX frame manager 303 programs the corresponding persistent bit to keep the frame marked as persistent. If the frame is re-enqueued into a persistent queue, then it will maintain its persistent status until deleted from the queue. Meanwhile, the transmit scheduler 307 operates as normal, dequeuing the next frame F 2 for delivery to the transmit function 309 for transmission. Upon successful completion of regular, non-persistent frames, the transmission scheduler 307 may return a completion status via the TX done queue 319 . In one embodiment, such completion status is not returned if the persistent frame was successfully transmitted and successfully re-enqueued. The persistent frame F 1 is repeatedly processed by the transmit scheduler 307 and resubmitted to the TX frame manager 303 via the re-enqueue path 321 . Operation repeats in this manner for as long as the frame F 1 is marked as persistent and the transmitter remains enabled. In one embodiment, a persistent frame is always resubmitted to the same transmit queue 305 from which it was retrieved. In an alternative embodiment, a persistent frame may be resubmitted to any of the transmit queues 305 by the TX frame manager 303 .

[0076] FIG. 6 is a simplified block diagram illustrating operation between the network I/O driver 219 and the MAC device 223 for clearing the persistent bit in a queued frame descriptor or a bit of the queue. In particular, the network I/O driver 219 submits a clear persistent (CLRP) command frame 601 along with a frame descriptor (FD) 603 that includes a frame pointer (FPtr) 605 descriptor number, or the like, that identifies or otherwise points to a particular persistent frame, such as the frame F 1 . The TX frame manager 303 includes clear persistence logic (CPL) 607 , which retrieves the frame pointer 605 from the CLRP command frame 601 to identify the particular persistent frame F 1 as shown at 609 . Upon identifying the persistent frame F 1 , the clear persistence logic 607 either modifies the PRST field of the frame descriptor or clears the persistent bit of the frame F 1 , as shown at 611 , so that it is no longer marked as persistent. In this manner, once the frame F 1 is no longer marked as persistent, it is processed in the same manner as a normal frame and not re-enqueued. In typical cases the CLRP command frame and descriptor is then discarded or marked as successfully completed and passed back to the network I/O driver 219 , but in any case the CLRP command frame is not transmitted.

[0077] The use of the PRST field to mark a frame as persistent provides many benefits and advantages for improving the control of wireless communications. In order to properly implement a packet oriented wireless communications protocol, and some types non-packet protocols, the application programs 218 and/or the network I/O driver 219 needs to conduct periodic functions or operations over predetermined or arbitrary periods of time. For example, one or more application programs at other stations on the wireless network may need to transmit successive frames of a voice stream with the transmissions occurring at predefined intervals corresponding to the sampling rate of their vocoders. This periodic service needs to occur with a high degree of uniformity of jitter of as little as a few tens of milliseconds can impair delivered audio quality. The best way to achieve this uniformity of transmission opportunity timing in a WLAN protocol like 802.11 is to use contention free frame delivery, in which the AP periodically polls the stations to facilitate such transmissions. For example, a WLAN may include three other wireless stations, W 1 , W 2 and W 3 , other than the access point controller that need to communicate. The access point main processor via the network I/O driver 219 periodically polls each of the wireless stations W 1 -W 3 in any particular order and according to any particular priority or predetermined service rate specification. For communications according to 802.11 for example, a CF polling list is maintained by the AP where one or more of the other wireless devices in the WLAN are periodically polled to enable communication with those devices.

[0078] In the conventional network interface card (NIC) model, such repetitive activities would require the scheduling entity 109 or the network I/O driver 219 to resubmit frames, such as CF polling frames for each repetition, although these types of activities are not as common for conventional, wired networks as they are for wireless networks. As described previously, however, the variable delay interface 105 imposes significant overhead and indeterminable delays on each such resubmission, which is an obstacle to the periodic functions being conducted in an orderly and repetitive fashion. During low levels of traffic, this requirement may be relatively easy to maintain. However, for periods of heavy traffic, and due to the variable delay interface 105 , it is difficult for host-based software, such as the scheduling entity 109 , to properly and timely perform the periodic functions.

[0079] The use of persistent frames facilitates the periodic functionality to occur without multiple or repetitive traversals of the variable delay interface 105 . The software, such as the scheduling entity 109 , need only mark one or more frames as persistent or enqueue the frames into a persistent queue or submit persistent frame types to establish the periodic retransmission of those frames to implement the periodic function. The persistent frames may thus be processed at the maximum rate allowed by the protocol rules and other, non-persistent frames, or may be synchronized with particular protocol-defined intervals by the combined use of the persistent and QM functions. The MAC device 223 automatically re-enqueues persistent frames after processing. The host system submits the clear persistent command to reprogram any persistent frame as a normal frame or to delete a frame from a persistent queue. In this manner, the persistent frame programmability enables the host system to control periodic functions, including polling frames, across the variable delay interface 105 .

[0080] FIGS. 7 A- 7 C illustrate the benefits of persistent frame capabilities for submitting polling lists employing polling frames marked as persistent. As shown in FIG. 7A, a selected one of the transmit queue 305 is loaded by the network I/O driver 219 with a polling list 701 including six CF-poll (“P”) frames PI-P 6 each marked as persistent. In this embodiment, six different wireless stations in the WLAN, such as wireless stations W 1 , W 2 , W 3 , W 4 , W 5 and W 6 , are each polled with a respective CF-poll frame P 1 , P 2 , P 3 , P 4 , P 5 and P 6 . After the wireless station W 1 is polled with CF-poll frame P 1 , the CF-poll frame P 1 is reenqueued to the corresponding transmit queue 305 by the transmit scheduler 307 and the TX frame manager 303 via the re-enqueue path 321 as previously shown in FIG. 5B . Since each of the CF-poll frames P 1 -P 6 are marked as persistent, the polling list order is maintained since each frame after being transmitted or otherwise processed is returned to the end of the same transmit queue 305 . As shown by the polling list 701 , the wireless stations W 1 -W 6 each receive an equal number of CF-poll frames per cycle through the polling list.

[0081] FIG. 7B illustrates an alternative polling list 703 loaded into a transmit queue 305 . In this case, the CF-poll frames are ordered P 1 P 2 , P 1 P 3 , P 1 , P 4 and so on. The frames of the polling list