Methods and apparatuses for a data packet flow control for recovering an excessive radio base station buffer dwell time

Description

TECHNICAL FIELD

[0001] This technology relates to packet data flow control in a radio communication system.

STATE OF THE ART

[0002] Contemporary cellular systems provide Internet browsing and access to services on the Internet. The packet data that the packet system communicates to the end user in these situations must be handled differently from circuit switched for speech in that the cellular system should maintain an unchanged packet flow between the internet source and end user so that services were not interrupted. Packet data flow control (simply referred to as flow control) with Wideband Code Division Multiple Access (WCDMA) technology helps to achieve this goal. Flow control in WCDMA is spread between two nodes: a radio network controller (RNC) and a radio base station (RBS). The RBS is responsible for maintaining a steady flow of packets to the user equipment (UE) over the air interface to provide a satisfactory user experience, e.g. while browsing the internet. Fig. 1 is a diagram used to explain WCDMA flow control.

[0003] The RNC is responsible for ensuring that data is available on the RBS for transmitting, and to enable this, the RNC transmits Radio Link Control (RLC) data via the TN Transport Network to the RBS. The RNC thus affects the buffer level in the RBS by transmitting RLC PDUs (RLC packet data units) to the RBS via the TN Frame Protocol (FP). The RBS sends the Bandwidth Grants (CA) as input to the RNC, and the RNC typically responds by sending an appropriate amount of data at a specific rate to the RBS via the or interface.

[0004] As the RLC protocol is terminated at the RNC and in the UE, any RLC data residing in the RBS queue is perceived to be delayed by the RLC entities involved in communicating the data between the RNC and the UE. As a result, the RNC, when the query timer expires, will query for data that has already been transferred from the RNC to the RBS, but which still resides in the RBS queue for the EU and is consequently not yet validated by the EU. Hence, if the amount of data in the Priority Queue RBS (PQ) (a term used in WCDMA) is large and the data rate over the air interface (Uu) is low, then the RLC entity in the RNC will query the RLC entity in the UE whether it has received the specified data packet even though no data is lost. Consequently, this will lead to unnecessary retransmissions of RLC packet data units (PDUs) with set query bit or super query fields (SUFI POLL).

[0005] Another problem is posed by RLC retransmissions. RLC retransmissions delayed by data already buffered in PQ can lead to multiple requests for the same data, even though previously retransmitted data is on its way but still buffered in PQ RBS. In this delayed retransmission situation, the UE may, due to the state of the timer, forbid the expiry or other status reporting trigger, send additional RLC status reports even before the initial

-1ΕΡ3 014 824Β1 the retransmission will be provided to the UE over the air interface, leading to a situation where multiple copies of the same data will be transmitted over both the TN and the Uu interface. This results in an inefficient use of TN and air interface resources as additional copies of the retransmitted RLC PDUs do not contribute to the perceived throughput of the user as they will be discarded as duplicates by the receiving RLC entity at the UE.

In traditional downlink flow control in WCDMA, i.e. for High Speed ​​Downlink Packet Access (HSDPA), the RNC arranges received downlink packets into a queue, hereinafter referred to as an RNC queue, and transmits the packets. to RBS in accordance with the maximum transmission bit rates via the interface or.

In more recent flow control approaches such as Active Queue Management (AQM) or Distributed Active Queue Management (D-AQM) flow control (Distributed Active Queue Management), RNC queue build-up is avoided by allowing RNC to transmit data to RBS as soon as possible. This results in queuing in the RBS and not in the RNC, unlike traditional WCDMA flow control which tries to throttle the queue length in the RBS to avoid spurious RLC retransmissions. As a result, it can be expected that the Ring Run Time (RTT) may vary to a greater extent for TN based on D-AQM flow control, as opposed to TN which uses traditional flow control. One way to deal with this would be to increase the RLC timers that are used to supervise the maximum allowable queue delays, but this would result in poor efficiency due to low peak throughput and excessive query delays in certain scenarios such as short data transmissions when lost. only the last RLC PDU (which carries the query).

[0008] While D-AQM tries to adjust the PQ buffer length in the RBS to the Uu rate, typically by targeting a certain buffer length or dwell time, the Uu rate for services over the air interface varies significantly in short periods of time, sometimes resulting in large PQ and low bit rate Uu. In these cases, the RBS takes time to shorten the PQ and adjust the lower Uu bit rate, leading to increased RTT RLC and another RLC query delay and retransmission. Consequently, D-AQM can have difficulties maintaining the desired queue length (PQ) which can lead to wasted bandwidth in both the TN and the radio network (RAN) due to unnecessary interrogation transmissions and multiple copies of RLC retransmissions.

[0009] One of the flow control tasks is to efficiently use the air interface between the RBS and the UE as radio spectrum is a scarce resource. If the flow of data packets is interrupted or unnecessarily slowed down, the user or the throughput of the cell will suffer. For this reason, it is important to keep RBS non-empty priority queues (PQs) with an adequate margin (a typical but example margin is 125 ms) to avoid poor throughput for the user.

[0010] Another important object of flow control is to avoid that the amount of data buffered in the PQ becomes too large and the residence time of the packet in PQ becomes too long. In other words, flow control should aim to keep the PQ in RBS as short as possible while ensuring that sufficient data is available to fully utilize the air interface capacity. In addition to the problems outlined above, RBS's large PQ lead to

-2ΕΡ3,014,824Β1 of the highest data loss with High Speed-Downlink Shared Channel (HS-DSCH) cell changes, which also leads to a large number of RLC retransmissions.

[0011] Yet another problem is that aggressive TCP applications do not always respond to traditional flow control schemes such as Active Queue Management (AQM) congestion control based on deliberate packet skipping. An example of an aggressive TCP application is one that starts a large number of TCP flows in parallel, each carrying only a small amount of data. In such scenarios, traditional AQM schemes may fail to keep the RBS buffer within target values, leading to an excessive build-up of the PQ buffer. The result is long download times as shown in Fig. 2, which is a graph showing an example of RBS buffer residence time when AQM flow control is used in a WCDMA system. When multiple data flows are suddenly started, there is an "uncontrolled" residence time in the AQM buffer as indicated in this figure. EP 1 603 359 A1 discloses a regenerative control method for recovering control of the flow of data packets between a network node and a radio base station, RBS, via a radio network interface. The RBS communicates with the user equipment, UE, via an air interface. The RBS receives measurement information regarding the data forwarding rate over the air interface from the RBS to the UE and the amount of downlink queue packets in the RBS associated with the UE communication. The process controller determines a commanded rate for forwarding data from the network node to the downlink packet queue based on the rate of forwarding data over the air interface and a factor determined according to the wait time during which the data packet is stored in the data queue. The network node forwards the data to the RBS according to the prescribed rate.

SUMMARY OF THE INVENTION

[0012] The technology in this application includes methods and apparatus that provide regenerative flow control for recovering control of the flow of data packets between a network node and a radio base station via a radio network interface.

[0013] The network node receives measurement information related to data transfer rate over the air interface from the radio base station to at least one UE and the number of downlink queue packets in the base station associated with the UE communication. The process controller in the network node processes the measurement information to determine the desired amount to be stored in the downlink packet queue. The process controller determines the commanded rate for forwarding data from the network node to the downlink packet queue based on the specified amount desired. The network node forwards the data to the radio base station according to the commanded rate. The steps of processing, specifying, and broadcasting can be repeated as required.

[0014] In one embodiment, obtaining the measurement information comprises receiving a message from the radio base station with some or all of the measurement information.

[0015] In another embodiment, the used data volume is calculated using the ordered rate, e.g. until the next message with measurement information is received from the radio base station. One or more data packets that exceed the calculated, used data volume are discarded from the buffer in the network node used to store the data packets for the data flow associated with the UE.

-3ΕΡ3 014 824Β1

[0016] In embodiments, the process controller performs a time-invariant linear process control to determine a commanded rate. The reference amount of data is determined using the reference residence time and the measured bit rate via the air interface. The determined reference amount of data is provided to the process controller, which processes the specified reference amount of data together with the measurement information using the invariant time-linear process control to determine the commanded bit rate. The prescribed bit rate is preferably limited to a non-negative value. The process controller preferably includes a first delay associated with the transmission of data packets from the network node to the radio base station and a second delay associated with signaling the measurement information from the radio base station to the network node. In one exemplary implementation, the process controller includes a feedback derivative-integrator controller and a feedback controller.

[0017] The process of regenerative control is distributed. The radio base station measures measurement information regarding data rate over the air interface from the radio base station to at least one UE and the number of downlink queue packets in the base station associated with the UE communication. The base station transmits the measurement information to the network node and then receives data from the network node at the commanded rate based on the measurement information such that the desired amount of data is stored in the downlink packet queue. Like a network node, a base station may repeat the measuring, transmitting, and receiving steps one or more times as needed. [0018] In one exemplary embodiment, a normal operation flow control process was used by the radio base station to control communication of data to the UE before the regenerative control method was activated. After the regenerative control method is disabled, the base station returns to the flow control process of normal operation.

[0019] In exemplary implementations, the regenerative control method may be activated when the residence time for data packets stored in the packet queue exceeds the first residence time threshold and turned off when the residence time for data packets stored in the downlink packet queue is less than the second dwell time threshold or the amount of data in the downlink packet queue is less than the data amount threshold.

[0020] In the exemplary implementation, the radio base station communicates with the UE using High Speed ​​Downlink Packet Access (HSDPA) technology, and the second flow control process is based on flow-based Active Queue Management (AQM) where the radio the base station determines a data rate over the air interface from the radio base station to at least one UE.

DESCRIPTION OF DRAWING FIGURES

[0021]

Fig. 1 is a diagram used to explain WCDMA flow control.

Fig. 2 is a graph showing an example of RBS buffer residence time when AQM flow control is used in a WCDMA system when multiple data flows are suddenly started;

Fig. 3 is a flowchart showing example procedures for a radio base station regarding regenerative flow control;

-4ΕΡ3 014 824Β1

Fig. 4 is a flowchart showing example procedures for a network node regarding regenerative flow control;

Fig. 5 is a function flowchart illustrating an exemplary communication system in which regenerative flow control technology may be used;

Fig. 6 is a function flowchart illustrating an exemplary communication system of the WCDMA HSDPA type in which regenerative flow control technology may be used;

Fig. 7 is a function flowchart illustrating an exemplary process control for implementing regenerative flow control according to non-limiting exemplary embodiments;

Fig. 8 is a function flowchart illustrating an example of the process control forward portion of Fig. 6;

Fig. 9 shows Bode plots of an exemplary portion of forward feedback (open loop) without driver, i.e. at Go (s) = 1;

Fig. 10 shows Bode plots for an exemplary derivative-integrator controller designed for a worst-case delay case, a crossover frequency of 0.65 Hz, and a phase margin of 55 degrees;

Fig. 11 is the Bode plots of the open-loop system of Fig. 9, with the derivative-integral control of Fig. 10;

Fig. 12 shows Bode plots of a feedback circuit (closed loop) with the derivative-integral control of Fig. 10;

Fig. 13 is a Nyquist diagram of a circuit with derivative-integral control;

Fig. 14 is a function flowchart illustrating an example of the process control feedback portion of Fig. 6;

Fig. 15a shows the frequency response of the forward feed controller of Fig. 14;

Fig. 16 is a block diagram of an example of a linearly connected feedback and forward feed controller;

Fig. 17 is plots of steered PQT for the lower delay case;

Fig. 18 is a graph of PQL controlled or buffer data volume / amount for the lower latency case;

Fig. 19 shows plots of steered PQT for the higher delay case;

Fig. 20 shows plots of PQL controlled or buffer data volume / amount for the higher latency case.

DESCRIPTION OF UNLIMITED EXAMPLES

[0022] The following section establishes special details such as specific embodiments for the purpose of explanation and without limitation. However, one skilled in the art will appreciate that in addition to these specific details, other embodiments may be used. In some cases, detailed descriptions of well-known methods, nodes, interfaces, circuits, and devices are omitted so as not to obscure the description with unnecessary detail. It will be appreciated by those skilled in the art that the functions described may be implemented at one or more nodes by application

-5ΕΡ3 014 824Β1 hardware systems (e.g., analog and / or discrete logic gates, interconnected to perform a special function, etc.) and / or using software and data applications in conjunction with one or more digital microprocessors or universal computers. Nodes that communicate using the air interface also have appropriate radio communication systems. In addition, the technology may further be contemplated for exemplary implementation entirely within any form of computer readable memory such as a monolithic memory, magnetic disk, or optical disk containing the appropriate set of computer instructions that would cause a processor to execute the techniques described herein.

[0023] A hardware implementation may include or include, without limitation, digital signal processor (DSP) hardware, a limited instruction set processor, hardware (e.g., digital or analog) circuits including but not limited to application specific digital circuits (ASICs), and / or user programmable gate circuits (FPGAs) and (where appropriate) state machines capable of performing such functions.

In terms of a computer implementation, a computer is generally understood to include one or more processors, or one or more drivers, and the terms computer, processor, and controller may be used interchangeably. When provided by a computer, processor, or driver, these functions may be provided by a single, dedicated computer or processor or driver, by a single shared computer or processor or driver, or by multiple individual computers or processors or drivers, some of which may be shared. either scattered. Moreover, the terms "processor" or "driver" also refer to other hardware capable of performing such functions and / or executing software, such as the exemplary hardware listed above.

[0025] It will be understood by those skilled in the art that "UE" is a non-limiting term, encompassing any wireless device or node equipped with an air interface allowing at least: transmitting signals on an UL (uplink) and receiving and / or measuring signals. on DL (downlink). Here, the UE may include a UE (in its general sense) capable of operating or at least performing measurements in one or more frequencies, carrier frequencies, component carriers, or frequency bands. This may be a "UE" operating in single or multi-RAT (radio access technology) or in a multi-standard mode.

[0026] A cell is associated with a base station, the base station generally including any downlink (DL) transmitting and / or uplink (UL) receiving node. Some exemplary base stations are eNodeB, eNB, Node B, macro / micro / pico radio base station, home eNodeB (also known as femto base station), repeater, repeater, sensor, radio transmitting only nodes, or radio receiving only nodes. The base station may operate or at least make measurements in one or more frequencies, carrier frequencies or frequency bands, and may be capable of aggregating carriers. It may also be a single radio access technology (RAT) node, multi-RAT, or multi-standard, e.g. using the same or different baseband modules for different RATs.

[0027] The described signaling occurs either over direct links or logical links (e.g., through higher layer protocols and / or through one or more network nodes. For example, signaling from the coordinating node may pass through another network node, e.g. a radio node.

-6ΕΡ3 014 824Β1

[0028] Exemplary embodiments are described in the context of a non-limiting example of a WCDMA-based radio cellular system. However, the technology is not limited thereto and may be applicable to any radio communication system including, for example, an Evolved UMTS Terrestrial Radio Access Network (E-UTRAN) LTE system.

[0029] The rule making and regenerative flow control technology in this application ensures that the buffer (queue) lengths of the radio base station are kept at acceptable values. Under base station conditions, the flow control routine in normal operation operates to control the flow of packets downlink to the UE for one or more data packet flows. Although flow control is applied to each data flow, only flow control for one data flow is described, for simplicity of description, regenerative flow control is triggered under predetermined conditions so as to avoid excessive buffer build-up in the radio base station buffer and is preferably removed. when the buffer fullness (i.e., the amount of data stored in the buffer for the data flow) reaches an acceptable level. First, the radio base station measures the residence time of the Priority Queue (PQT) packets in the PQ. If the PQT is equal to or less than a certain threshold, no specific action is taken. However, if the PQT exceeds this threshold, the following actions are taken. The radio base station measures (1) the amount of data stored in the base station buffer for the data flow, also referred to as Priority Queue Length (PQL) and (2) the transmitted data rate from the base station to the UE associated with the same data flow. RBS compares PQL and PQT to a first set of relevant predetermined threshold values ​​corresponding to the maximum allowable PQL and the maximum allowable PQT. If both the specified PQL and PQT exceed their respective threshold values, then the radio base station transmits the measurement information to the core or radio network controller node (RNC) via uplink signaling included in the uplink frame protocol. In a non-limiting example of WCDMA, an RNC is used as an exemplary network node. The RNC uses the data rate and PQL from the measurement message received from the base station to calculate the Priority Queue Time (PQT). Receiving this information triggers the RNC to initiate recovery flow control. A second set of relevant predetermined thresholds is used to determine when to deactivate regenerative flow control and return to normal flow control, e.g. flow control written only on the radio base station. Consequently, the first and second sets of thresholds define the ranges within which regenerative flow control is active and limits the data flow from the RNC to the RBS. Thus, unlike known flow control techniques, where only RBS alone determines the rate requested from the RNC via bandwidth allocation control frame messages, the regenerative flow control technology in this application is used selectively under certain conditions. The interaction between RNC and RBS in regenerative flow control technology dissipates regenerative flow control between RNC and RBS.

[0030] Some preliminary definitions are provided to assist the description, but are not intended to be limiting. For example, the concept of a priority queue is used with the understanding that the priority queue includes other concepts for a data store such as queue, buffer, memory, etc. which is used to temporarily store data at a base station before transmitting over an interface.

-7ΕΡ3 014 824Β1 radio.

[0031] PQL - Packet Queue Length - Current volume or amount of data measured in bits stored in a Priority Queue (PQ) in the RBS.

[0032] Rate Uu - Current data rate measured in bits / sec at which data bits leave PQ for transmission over the air interface from the RBS to the UE.

[0033] PQT - Packet Queue Time - The amount of time, measured in seconds, that an incoming data packet in PQ will spend, given the current PQ data volume and the current radio transmission rate (Uu) from the RBS to the UE. PQT is also known as residence time.

[0034] Accordingly, the distributed method and apparatus perform regenerative control when the PQT becomes too long, and when the PQT returns to normal, flow control is switched back to normal flow control, e.g., AQM based on flow control. some WCDMA systems.

[0035] Fig. 3 is a flowchart illustrating exemplary procedures for a radio base station regarding regenerative flow control according to an exemplary embodiment. In Step S1, RBS measures the current PQT (Time spent in the packet queue) in the RBS. If PQT does not exceed the first residence time threshold 1 (PQT), then the control returns to step S1. If the PQT exceeds the first residence time threshold 1 (PQT) (step S2), then the regenerative control process is activated and the RBS starts measuring the current PQ data volume (PQL) which is the size of the downlink packet queue at the base station associated with the communication. from the UE, and the current rate Uu from the RBS to the UE, i.e. the current rate for communicating data over the radio interface from the radio base station to the UE (step S3). The RBS signals the measured PQL, PQT and bit rate Uu to the RNC (step S4). The RBS receives data from the network node at the commanded rate received from the network node (described below) in step S5. The RBS then checks to see if PQT <threshold 2 PQT (second dwell time threshold) or that the current data volume PQ (PQL) <threshold 3 PQL (threshold amount of data) (step S5). If any of the conditions are met then the RBS reverts to normal flow control and monitors the PQT at step S1. If neither condition is met, then RBS continues regenerative flow control and returns to step S3.

[0036] Fig. 4 is a flowchart that shows exemplary network node procedures followed by an RNC as a network node in this example. The RNC obtains some measurement information (step S10) of the downlink packet queue size at the base station associated with the communication with the UE, i.e. the data volume PQ or PQL, and the data transfer rate over the air interface from the radio base station to the UE, i.e. the rate Uu. Residence time, i.e. PQT, may be determined using a rate of PQT = PQL / Uu. In one embodiment, measurement information, e.g., PQL and bit rate Uu, may be measured by and received from a radio base station. In another embodiment, the RNC may use the data rate information already available at the network node, and / or may use the current Radio Link Control Window (RLC) size to estimate the PQ data volume, rather than receiving one or both of the rate and volume information from a base station. . When an RNC receives a measurement information message from a base station, receiving such a message may activate a regenerative flow control process in the RNC. However, other factors that trigger the regeneration process may be used. The RNC processes measurement information to determine the desired one

-8ΕΡ3 014 824Β1 amount of data to be stored in PQ (step S11). The RNC determines the ordered rate based on the specific desired buffer size to be used for transmitting data from the RNC to the RBS via the interface or (step S12).

[0037] In the embodiment detailed below, this commanded data rate determination is performed using a process control based controller algorithm, sometimes referred to as controller servo algorithm. In embodiments, the controller servo algorithm is a derivative-integral controller designed to guarantee stability within a predetermined range of signaling and control delays in the control loop between the RNC and the RBS. The RNC calculates the used data volume using the commanded unit of the rate at which the next measurement information message is received from the base station (step S13). The RNC may then discard from the buffer in the RNC to the data flow, data packets exceeding the specified buffer size (step S14). For example, packets may be dropped from one end, like the top of the RNC buffer. Then, the RNC queues the transmission of the SDU from the RNC queue to the RBS at the commanded rate (step S15). In step S16, a decision is made as to whether a new measurement report is received from the RBS. If so, then the RNC returns to step S11 to calculate a new commanded rate based on the new measurement information. If not, then the RNC (RBS) reverts to normal flow control, e.g. AQM flow control (step S17).

[0038] Fig. 5 is a function flowchart illustrating an exemplary communication system in which regenerative flow control technology may be used. Network node 14, e.g. a core node or RNC, is connected to one or more networks 1 such as the Internet, etc. via network interface 50 to one or more Radio Base Stations (RBS) 18 via an RBS interface 60. Network node 14 includes one or more data processors 52 coupled to interfaces 50 and 60, as well as one or more memories 54 that stores program instructions 56 and data 58 necessary to implement the flow control operations described in this application. RBS 18 communicates via network interface 2 with network node 14 via network node interface 62. RBS 18 also communicates with one or more UEs 10 over a radio interface, via an air interface 72 which includes, for example, one or more radio transmitters, receivers, antennas, etc. RBS 18 further includes one or more data processor 64 connected to the interfaces. 62 and 72, as well as one or more memories 66 that stores program instructions 68 and data 70 necessary to implement the regenerative flow control actions described in this application.

[0039] Fig. 6 is a function flowchart illustrating an exemplary WCDMA HSDPA type communication system in which regenerative flow control technology may be used. RNC 14 is coupled via a transport network (TN) 16 to a node of the radio base station 18 to forward the data packets from the end devices 12 to the UE 10 via the air interface 25. Transport network 16 is in this example an or transport network, and may include one or more transport nodes (not shown) that are coupled so as to allow data packets to be sent between RNC 14 and RBS 18. Terminal devices 12 may correspond to various types of devices, including servers, remote terminals, and further UEs. In addition, the further control node is coupled between the RNC 14 and the end devices 12 and corresponds to a Serving General Packet Radio Service Support Node (SGSN) 16.

-9ΕΡ3 014 824Β1

[0040] As shown, different protocols or protocol layers may be used to transmit the data packets between the end devices and the UE. Fig. 6 shows an end-to-end bonding protocol implemented between end devices and UEs, a retransmission protocol implemented between the RNC 14 and the UEs and a flow control protocol implemented between RNC 14 and RBS 18. The end-to-end protocol is a higher layer protocol, and in the illustrated example, it could be TCP / IP (TCP: Transport Control Protocol, IP: Internet Protocol). Other types of protocols may also be used for rate control, e.g. UDP (User Datagram Protocol) together with the Real Time Protocol (RTP) and / or its equivalent, the Real Time Control Protocol (RTCP). The retransmission protocol is a lower layer protocol, and in the example shown is the RLC protocol, which is a linker layer protocol. According to the retransmission protocol, unsuccessfully received data packets are retransmitted which involves some type of feedback mechanism from the retransmission protocol receiver to the retransmission protocol sender, e.g. by forwarding acknowledgment packets from the receiver to the transmitter. Fig. 6 also indicates radio access carrier channels established between the SGSN and the UEs. The carrier channel is considered to be a channel having some guaranteed transport attributes, e.g. with respect to the Quality of Service (QoS).

[0041] RBS 18 includes priority queues (PQs) 22, measurement controllers 24 and 26 for measuring the data rate on the air interface to the UE and the current volume in the associated PQ 22. The overflow detector 26 detects if the PQ has too high a residence time for the associated packet, and in this case, the measurements of the bit rate Uu 24 and the data volume 26 are initiated. The message generator 30 generates a measurement message when the conditions for regenerating a predetermined flow, such as those described in connection with Fig. 3, are met. The measurement message includes measurement information associated with the current data flow PQL, PQT and Uu rates. Message transmitter 32 transmits a measurement message from message generator 30 via TN 34 to RNC 14.

[0042] RNC 15 includes packet queues 44 to store data for the UEs received from the SGSN 16. Message receiver 36 receives measurement messages transmitted via TN 34 from message transmitter 32 and triggers regeneration activator 38 to run regenerative flow control for that data flow. Regenerative flow controller 40 performs regenerative flow control for this flow until deactivated by the regeneration activator 38 when the message receiver 36 detects that measurements communicated from the base station message transmitter 32 have been stopped. The flow controller 40 sends the commanded data rate determined from the measurement information using a process control algorithm as in the example described below to packet transmitter 42. Packet transmitter 42 transmits packets to RBS 18 at the commanded data rate during regenerative flow control for the corresponding data flow. When measurements from message transmitter 32 are stopped, normal flow controller 23 performs normal flow control.

[0043] The inventors have developed a regenerative flow controller in the context of a process control or servo-control problem, where a steered signal is applied to a reference signal. They further determined that the regenerative flow controller should also handle a number of restrictions

-10ΕΡ3 014 824Β1 of controlled system, i.e. PQ in RBS, which are associated with (1) measurements performed in RBS, (2) delays associated with signaling these measurements from RBS to RNC, (3) throughput effect determined by process control from RNC to RBS and (4) the need for the commanded rate determined by the process control to be non-negative.

[0044] Restriction (1) refers to the risk that the RBS measurements are subject to some degree of fluctuation (variation in the occurrence of sampling time). The delay constraints (2) are related to the signaling and the effect of the rate determination by the process control. These delays are not known when designing the regenerative flow controller, are variable in time and not measurable for the time required by the regenerative controller, and have a wide range, e.g., from 10 ms to 130 ms. In fact, the effective sampling period delay, measurement delays and commanded bandwidth signal delay add up, which means that the regenerative controller has to handle an unknown delay. Using the non-limiting example delays, the summed delay can be from 10ms + 10ms + 20ms = 40ms, up to 130ms + 130ms + 20ms = 280ms. The term "handle" means i) the performance in terms of the time to settle to the desired reference signal must be as specified for the delays in the above range, and ii) that the stability of the driver is guaranteed for all these delays. The restriction (4) means that any negative control signal must be replaced with a non-negative value. This can adversely affect the regenerative controller so that some mitigation may be desired.

[0045] The inventors have recognized that the delays are not known and cannot be measured with high bandwidth, and that there may be fluctuations. In addition, discrete time control relies on accurate sampling or potentially on-line interpolation of measurement. As the former cannot be guaranteed, and the latter increases the complexity, the exemplary embodiments below use a continuous-time construction. The delays, jitter and sampling are combined into an effective delay, and then a continuous time controller is designed. This continuous-time controller is discretized and operated as if the measurements were not subject to an effective delay as will be further described below.

[0046] As delays are unknown and cannot be measured, the regenerative controller is designed to be resistant to changes in delay. Since delays cause an increase in phase, the feedback portion of the exemplary regenerative controller is designed for the greatest possible delay and to ensure stability and efficiency for lower delays.

[0047] As mentioned, the purpose of the process control is to direct the PQT towards the reference PQT value. There is a closed-loop control portion of the regenerative controller that is based on the output measured with PQ. How PQT and PQL are used in this context will be described below. The current uplink data rate for the data flow over the air interface (bit rate Uu) is also measured. The current uplink data rate over the air interface has a direct effect on the PQ, but cannot be manipulated by the regenerative controller. In automatic process control technology, such a signal is classified as a measurable disturbance that can improve operational efficiency through the use of feed-forward or open-loop process control.

[0048] The controlled system is preferably designed for linear control. One of the benefits of this linearity

-11ΕΡ3 014 824Β1 is that the principle of superposition allows the feedback and feedback parts of the controller to be designed separately. These separate parts are then linked together to form a combined feedback and forward feed controller.

[0049] Fig. 7 is a function flowchart illustrating an exemplary process control for implementing regenerative flow control according to non-limiting exemplary embodiments. The process control output is the commanded transmit data rate for the data flow from the RNC to the base station. The commanded bit rate is converted to the data volume PQ by integration as shown in integrator block 90 in Fig. 7. Integration is linear and time invariant and PQT is derived from the equation

PQT_and(t) = ^^, (1) where ^ is the PQT of flow i at time t, U is the data volume of flow PQ i at time t, and where LUU ί Λ M

V / is the bit rate Uu of flow i time t. This means that the feedback from PQTi (t) would have to be based on a time-varying continuous-time model, which the authors decided was less desirable than a linear time invariant model. The inventors overcame this problem by embedding a time-invariant linear driver design into a time-varying linear driver. This embedding allows feedback control to be applied based on the measured U (pql), rather than based on the measured (^pQT) use of the time-computed reference variable vm ref^PQT data volume PO ^ / e ^ Instead of reference PQT ^J.'

-br (i) U)

[0050] The diagram in Figure 7 describes a process controller and control system for one data packet flow i, but the technology is applicable to multiple flows. The figure shows that the controllable system and controller are distributed between the RBS and the RNC and that the data and signaling used are_rpfPQT is the interface or. The PQT 'reference value is first multiplied in the multiplier 80 by measuring m (. _ ΓΤΝ, υΐ.). .

the current bit rate Uu^{D |} '' 'to create a reference value for the PQ / w data volume^f'^rThis operation and the regenerative control algorithm implemented in the process controller 82 occurring in the RNC. Process controller 82 receives the reference value of the data volume PQ currently measured Κ Ι-Ή h ^ G- ^ OK · · X Hf · fh data volume ^ '^v ' ⁷and the current measured bit rate 00⁷¹¹ Based on these inputs and the internally stored states described below, controller 82 computes a control signal ^{Ł /}X^rwhich is a commanded rate to be used in transmitting data from the RNC to the RBS during regenerative flow control.

[0051] In Fig. 7, RBS is denoted by NodeB according to the 3GPP standard. The commanded bit rate is then used to schedule data transfer via or. This transmission uses a transport network (TN). Depending on what kind of TN is used and the load on TN, it occurs

-12ΕΡ3 014 824Β1 πΤΝ, ΟΙ.

delay¹ and before the transmitted data stream arrives at the RNC and starts adding itself to the PQ.

.j? · ™ "

This delay is represented by the corresponding Laplace transform e '' 84 in Fig. 7.

Similarly, the signaling delay associated with the two measurements is represented by - \ T^him'Laplace transform e' 94, 96 towards UL (uplink). According to e ', delay 84 is a limiter 96 limiting the commanded bit rate to a positive value.

[0052] Node B (RBS) implements PQ as an integrator 90, represented by a Laplace transform of 1 / s.

O- ·· Η I - hH C (, -7 ^) = ^) - I -_wrf·

Inbound traffic difference <\ / '^v'and the air interface rate Uu "E is integrated to give the data volume PQ ^ <V / -PQT (residence time) is obtained by dividing by (She a factor of 92.

[0053] In embodiments, a differential integral controller is used for the feedback portion of the servo driver for regenerative flow control. Fig. 8 is a function flowchart illustrating an example of the process control feedback portion of Fig. 6. To describe the control system used to design a feedback controller C (s), where s is a complex variable of the Laplace transform, the following equation is useful. :

<img file="PL3014824T3_D0001.tif" />

(3)

G._ABOUT0 = G, 06₂0C0D0, (4)

In (3) and (4), ^G.&lt;/^s'means open loop system, ^³'is the data volume (PQL) and -'is the desired data volume. The Go (s) factors are the transmittance delay carry function Gi (s), the signaling delay carry function D2 (s), and the PQ transfer function G₂(s), These quantities are explained above and given by equations (5) - (7).

G._}{s) = e (5)

C.₂0 = -(6)

D (s ^ e - ^ (7)

[0054] The transfer function of the driver C (s) is designed to be the derivative-integral driver here. The transfer function of this driver structure is given by:

= (8) as + bN s + - · M where driver parameters K, a, M&gt; 1, b and N&gt; 1 are determined by standard methods to fulfill control tasks (settling time and stability).

-13ΕΡ3 014 824Β1

[0055] A standard design procedure can be outlined as follows. First, based on the requirement to meet the reference value with some steady state accuracy, the desired level of integral control in the controller is selected by selecting M. The higher the value of M, the more integral action will follow. A value close to 10 is used as an example. Second, the parameter a is chosen to be significantly less than the crossover frequency, e.g., ⁹ “Py to limit the phase loss at the crossover frequency. Third, based on the steady-state time, the preferred cross-over frequency is determined, which is defined as the frequency, where

- ¹ The result of this selection is a closed-loop feedback bandwidth close to the selected crossover frequency. This means that the parameters of the driver are selected such that this limitation is met. Fourth, based on the stability requirement, the desired phase margin is determined so that the phase at the crossover frequency is⁺

An example value might be = &lt;P_martand_n+ - tan (&lt;o_cM / a) - tan (aja) π to compensate for the phase loss of the derivative link. is a nominal phase margin, e.g. 60 °. Fifth, the phase advance to be provided by the integrating link is computed so that arg (G_about(/ o&gt;_c)) - -180 + 9% arg / n · This means that the sum of the phases is determined

180 &lt;p._um = --- (argfo (/ A)) + arg (G₂ (I_c)) + arg (p (; *&gt;_c))) ^π followed by calculation of required phase advance as Piance '180- &lt;p_SUfn + 9&gt;_hasrg w Then, the derivative link parameters are ^{1 + 2sin}

<img file="PL3014824T3_D0002.tif" />

। βθ Φadyance

<img file="PL3014824T3_D0003.tif" />

SOMETHING

<img file="PL3014824T3_D0004.tif" />

<img file="PL3014824T3_D0005.tif" />

computed as

<img file="PL3014824T3_D0006.tif" />

J ω² + (a / M)²

[0056] The gain K is determined to give l^G.o (/ ®c) l ¹to meet the definition of frequency + a²A - demarcating. This means that

[0057] A non-limiting exemplary design is shown in figures 9-13. Fig. 9 shows Bode plots of an exemplary portion of the forward feedback (open loop) without driver, i.e. with Go (s) = 1 and with no compensation, i.e. C (s) is set to 1. It can be seen that the amplitude curve is the same for all delays, represented by the absolute value of G2 (s). The phase curves are different, with the highest phase loss occurring for the case with the maximum total loop delay.

[0058] Fig. 10 shows Bode plots for an example derivative-integrator controller,. 14.

ΕΡ3 014 824Β1 designed for a worst case delay, a crossover frequency of 0.65 Hz, and a phase margin of 55 degrees. It has a high gain for low frequency which is a consequence of selecting M. This selection corresponds to a leaky integrator in the controller ensuring that the steady error becomes small enough. Phase shift peaks at the selected crossover frequency of 0.55 Hz.

[0059] Fig. 11 shows Bode plots of the open-loop system of Fig. 9, with the derivative integral controller of Fig. 10. Note that the phase is clearly above -180 degrees at the cross-over frequency due to the introduced phase advance.

[0060] Fig. 12 shows Bode plots of a feedback (closed loop) circuit with the derivative-integral controller of Fig. 10. There are some moderate resonance peaks that correspond to repeating poles introduced by the loop delay. However, they are not serious and the driver is expected to behave well. This is confirmed by the Nyquist plot of Fig. 13. Since all curves are to the right of -1, the closed-loop system is stable.

[0061] The process controller may be implemented on one or more computers. In this case, the continuous-time construction must be converted to discrete time using a discretization scheme. One example diagram is Tustin's method where the discrete time difference equation is obtained from the continuous time difference equation by substituting

1-^¹S -&gt; - r.

T \ + q

Here, T is the sampling period (typically 40 ms in the current application) and qr¹ stands for the backshift operator ~

[0062] By inserting (9) into (8) and applying the control error ^ (0 = (10)

This gives the following difference equation for the driver calculation:

0) ~ 0^— 7) + rp, 2^atί, β 0 ^— 27) + s ^, _vole (t - ^) + (11)

[0063] The driver parameters are functions of the continuous time and sampling period parameters via (9) and are not reproduced here.

[0064] Fig. 14 is a function flowchart illustrating an example of the feed forward portion of the process control of Fig. 6. A standard calculation using the linearity of the flowchart of Fig. 14 shows that the output of the forward feed portion of the controller can be written as:

y, (s) - G, (4g, (s) F, (, t) D (s) -1&gt;, "(.r). (12)

Here y {s} is again the data volume PQ (POL) and ^ 'M is the (measurable but unregulated) flow rate i. Note that the superposition principle allows the feedback driver to be neglected when designing the feedforward portion. The object of the feedforward portion is to cancel the output distortion effect; hence, it is desirable that (12) be close to zero. Ideally this is obtained if the forward feed function of the feedback driver satisfies the condition

-15ΕΡ3 014 824Β1

F, (s) =,.

Putting (5) and (7) into (13) gives you (13) (14)

[0065] Equation (14) represents the exact time shift of the signal, i.e. the feedback controller is non-causal and unworkable. Therefore an approximation is needed. First-order Taylor series expansion gives

F._f (s) »1 + + T ™ '^DL}, (15) which replaces the exact time shift forward in the predicted time forward. Now, as stated in (15), a differentiation of the signal (s) is introduced, which is undesirable since the high frequency noise would be amplified. Therefore, a final adjustment is made where (15) is lowpass filtered as

F._f U) * (1 + + T ™ -^DL )) k (s), (16) where L (s) is the low-pass filter. In a preferred embodiment, a second order filter is used, i.e.

o&gt;²₀ s² + 2ζίυ₀ + ά&gt; ο

In (17), 'is the attenuation of the low pass filter, a is the bandwidth in terms of the angular resonant frequency of the second order filter.

[0066] The construction example appears in Fig. 15 in terms of the Bode plot. The feed-forward driver example is designed for a bandwidth of 4 Hz. As can be seen from the phase curve, the controller introduces a phase advance just like a derivative link. In addition, the feedback controller is almost 10 times faster than the feedback controller, and therefore can handle high frequency parts that the feedback controller will not handle.

[0067] And again the discretization is performed by the method of Tustin (9) and the resulting feed-forward regulator takes the form

- T)₊^₂at_{l jr}(t - 2T)₊s ^ bf (r)₊s ^ b ™ (tT)₊ s ^ b? (t - 2T) (14)

[0068] The driver parameters are functions of the continuous time and sampling period parameters via (9) and are not reproduced here.

[0069] The feedback controller, the feed-forward controller and circuitry have been assumed to be linear so far. In this state of affairs, the principle of superposition applies, and the feed-back and feed-forward effects can be added together to form a complete servo driver.

[0070] Fig. 16 is a block diagram of an example of a linearly connected feedback and forward feed controller. The diagram is the sum of Figs. 7 and 12. The control signal is a sum of feedback control and feed- forward control signals.

[0071] One further non-linear addition is needed. The linear driver can create negative control commands during the transition phase, producing a negative bit rate from RNC to NodeB, i.e.

-16ΕΡ3 014 824Β1 data in reverse. To avoid this, the following was introduced:

«= Max (« ™ u (15) where u_min S 0 is a preconfigured low minimum bit rate.

[0072] To test the efficiency of the controller operation, regenerative control was tested from an initially high PQ and PQT data volume by way of a non-limiting example. The simulation result appears in Figures 17-20. It can be seen that the regenerative flow controller directs the PQT to the desired 0.125 ms value in all cases. As expected, the control oscillates to some extent in the event of a long delay.

[0073] There are many advantages to the regenerative flow control technology described in this application. One exemplary advantage is that this technology ensures that excessive RBS PQ buffers are avoided while ensuring that sufficient data is still available to fully utilize the radio interface bandwidth. Another exemplary advantage is that the regenerative flow control mechanism is flexible in that the setting of the thresholds and the rate at which feedback information is transmitted can be adjusted to suit different feedback delays and target buffer conditions. A further exemplary advantage includes: reduced risk of excessive PQ buffers, which in turn maximizes air interface utilization and maintains high throughput; limited need for buffer memory in RBS; and improved throughput and end-user experience. In terms of flexibility of application, the technology described in this application is applicable to any other radio system that may use distributed queuing.

[0074] While the foregoing description is detailed, these should not be construed as limiting, but merely illustrating some presently preferred embodiments. The embodiments described herein may be taken as independent embodiments, or may be considered in any combination with each other to describe non-limiting examples. Although non-limiting, exemplary embodiments of this technology have been described in the context of WCDMA, the principles of the described technology can also be applied to other radio access technologies. Indeed, this technology fully encompasses other embodiments that may be apparent to those skilled in the art. Reference to an element in the singular is not intended to mean "one and only one", unless expressly stated so, but rather "one or more". Moreover, it is not necessary for a device or method to address each and every problem the solution of which is sought by the described technology to be covered by it.