Patent Application
20070127513
The present application may relate to radio communications systems and more particularly to systems and methods for managing the servicing of multiple data streams in a mobile radio environment.
Technologies are being developed to support data services in mobile environments such as the General Packet Radio Service (GPRS) and Universal Mobile Telecommunications System (UMTS) among other communication protocols. One example of these data services relates to connection through the Internet. Most popular Internet applications such as the World Wide Web (www), file transfer protocol (ftp), and e-mail use the reliable transport service provided by the Transmission Control Protocol (TCP). The performance perceived by users of such applications depends on the performance of the TCP. TCP has been designed and adapted to perform well in traditional wired networks, where data packet losses are assumed to be a symptom of congestion rather than failure in the transmission processes, and where the transmission bandwidth is often sufficient to support a multiplicity of simultaneous applications. Wireless communications systems, however, suffer from handover disconnections, corruption losses, and low and variable bandwidth, which dramatically affects TCP performance.
Transmission Control Protocol (TCP) belongs to the Transport layer of the TCP/IP networking protocol suite. TCP is a connection-oriented protocol that provides a reliable, full-duplex, byte stream for user applications. TCP uses a sliding window protocol for transmitting/receiving data with its peer. The peer is any other device with which communication is being attempted or conducted, at this layer of the protocol, and which has a unique internet protocol (IP) address and port number. A sliding window is based on the advertised “Window Size” of its peer, which is the number of data packets that can be sent without waiting for an acknowledgement (ACK). If there is no congestion in the network, then at anytime after TCP established a connection and determined the size of the appropriate congestion window, the amount of data that TCP has transmitted and waiting for ACK is equal to the window size of the peer or the congestion window, whichever is the lesser.
One problem which arises in a mobile terminal, also referred to as “user equipment” (UE), is starvation of the TCP/IP applications in a multi-application situation due to head-of-the-line (HOL) blocking. The rate at which the TCP/IP data that is queued in a UMTS stack is transmitted out of the UE depends on the data transmission capacity of the physical connection. For example, the radio link between the UE and a base station. In the event of multiple connections or sessions sharing a wireless link, first-in first-out (FIFO) packet scheduling can cause a HOL blocking effect resulting in unfair sharing of bandwidth and effects which may be observable by the human user.
The RLC sublayer 14 provides Automatic Repeat Request (ARQ) functionality, transparent data transfer, unacknowledged data transfer, acknowledged data transfer and buffer occupancy. The PDCP 16 functions include: compression in the transmitting entity and decompression in the receiving entity of redundant control information and may include TCP/IP header compression and decompression. Additionally, the PDCP may add control information for each data packet by adding a header.
Exemplary embodiments may be better understood with reference to the drawings, but these embodiments are not intended to be limiting. Like numbered elements in the same or different drawings perform equivalent functions.
When a specific feature, structure, or characteristic is described in connection with an embodiment, it will be understood that one skilled in the art may effect such feature, structure, or characteristic in connection with other embodiments, whether or not explicitly stated herein.
Embodiments may be implemented in hardware, firmware, software, or any combination thereof, and may include instructions stored on a machine-readable medium, which may be read and executed by one or more processors. Embodiments are disclosed in the context of a plurality of applications in a client, which may be a mobile communications device, such as a cellular radio, text messaging device or the like, and the protocols are exemplified by, for example, TPC/IP transport and network layer protocols, for convenience, however it is not intended that this be a limitation. Applications mean any software, firmware or hardware implemented functions communicating directly or indirectly with another application, a client or a server. Such applications may include text messaging, audio or video, e-mail and the like, as well as housekeeping functions such as status reporting. Wireless communication means may include, audio, radio, lightwave or other technique not requiring a physical connection between a transmitting device and a corresponding receiving device. While the communication is described as being from a transmitter to a receiver, this does not exclude the reverse path, and a wireless communications device may include both transmitting and receiving functions.
The terms “data queue”, and “buffer” are generally used interchangeably to refer to a one or more readable data elements stored in memory and used to represent the information being processed or being transmitted by an application, a program, an operating system or the like. Buffers may be of any size as measured in computer readable information, and be established, managed and eliminated under the control of a stored program or other similar mechanism, including firmware or hardware.
As shown in
In accordance with the TCP/IP protocol, each active application having a data stream may be represented by a 4-tuple that describes the session connection. At the UE 22, the 4-tuple includes a port number at each of the UE and the peer (source port, destination port). Port numbers may be assigned on an ephemeral or static basis for each application data stream. In certain circumstances applications may be assigned more than one port number. The 4-tuple also includes the IP address of the UE and the peer. The server 32N, or peer, of the UE 22 is reached by traversing a telecommunications network, which may be the Internet 30, after being transmitted over a wireless link 46. At the server 32N, a destination port is associated with the server application being connected to, and is often what is termed a “well known” port number, being designated for convenience in specifying the service required of the server by the UE application (client). The combination of the client port number and client IP address (“first socket”) and the server port number and server IP address (“second socket”) defines a data flow, data stream, session or connection.
In an embodiment, the functionality of the PDCP sublayer of the second protocol layer may be as shown in
A local flow control 56 is established between the PDCP 52 and the RLC 54. The flow control may be maintained such that the quantity of data in the RLC queue is sufficiently small for the process to be responsive to changes in user application requirements and the activation of new user applications, but sufficiently large as to prevent data starvation at the physical layer 10.
For each active data flow, a queue Qk is dynamically created to store the application data dispatched by the network and transport layers (TCP/IP). Each queue Qk may be established when data associated with unique client port number and server data port number and IP address arrives at the PDCP 52 and there is no preexisting queue for the specific data flow. The queue Qk for the data flow exists until all of the data has been unloaded from the queue. Where the application is transmitting more data than can be accommodated in the associated queue, the queue may be continuously reloaded by the application and TPC/IP protocol such that the queue is maintained until the desired block of data has been transmitted. Alternatively, a fixed queue size is established, and the queue is extinguished when the data has been transmitted. Further data in the data stream results in establishment of a new queue. Where multiple applications are active, there are multiple data flows and therefore multiple queues simultaneously extant.
Data packets are de-queued from each of the queues in accordance with a scheduling algorithm in the scheduler 48, which may be a round-robin scheduler. The size of the data packet may be varied, and either one packet or a multiple of packets de-queued by the scheduler. The de-queued packets are dispatched from the queue Qk to the RLC 54 after an appropriate PDCP header has been applied. Header compression 18 may also be performed.
In an aspect, a packet of data may be the maximum of the actual data to be transmitted and the MTU (maximum transmission unit) of the physical layer 10, including any header and trailer information required by the data protocols. In another aspect, the queue size of each queue associated with a data flow may be set by the maximum of the peer window size or the congestion buffer, whichever is less.
The operation of a round-robin scheduler 48, for a situation where there are two active data steams having associated queues, results in quantity of data (a data block), as described above, being removed from the first queue and dispatched to the RLC 54. The second block to be dispatched is from the second queue, representing the second data stream. The third block is then removed from the first queue. The amount of data remaining in each of the queues may determined by counting the quantity of data being removed from the queue during each execution cycle of the scheduling algorithm and the quantity of data added to the queue during the execution cycle of the scheduler 48, and adjusting the queue data quantity by the difference therebetween. Additionally, the number of active queues QN is determined periodically, so that new queues may be added to those being serviced by the scheduler 48, and empty queues deleted from the servicing process. As the location of queues may be in random access memory 42, the size and location of such queues is determined by an allocation table or other management means, such as memory address pointers. Determination of queue size, quantity and identity may be done on a time periodic basis, a fixed number of scheduler cycles, or multiple times per scheduler cycle. Applications may also be configured to indicate the total amount of data remaining locally to each application which has not as yet been transferred to a queue in the PDCP 52. This may be useful in reporting the total data transmission backlog of the UE when a large amount of data remains to be transmitted.
A round robin scheduler adapts to the number of active queues, representing active data streams, by dispatching a packet from each one of the active queues in turn until all of the queues have been accessed. The sequence is repeated, taking account of the addition and deletion of queues which may have occurred in the interim. Queues may be added or deleted dynamically during each round robin cycle, or at the conclusion of a first round robin cycle before initiation of a second round robin cycle. This process may be performed each time the RLC 54 requests data, and may result in filling the RLC 54 buffer. One round robin cycle may be performed each time the RLC 54 requests data 56; however this may not provide enough data to keep the RLC buffer fully loaded. Consequently, when the RLC 54 requests data the round robin process may be performed until the RLC buffer has been fully loaded, or until a quantity of data specified by the RLC request 56 is loaded into the RLC buffer.
In addition, the RLC 54 may request data from the PDCP 52 to determine the present data loading of the applications. In this example, the PDCP 52 reports the sum of the contents of the existing queues Qn, taking account of the compression and formatting of the data and PDCP headers, and may also include the quantity of data reported by the applications for each active data stream. This may permit the establishment of a local flow control loop with the PDCP 52 with the requests for data 56 from the PDCP layer being regulated by the data transmission capacity of the physical layer 10. In reporting the loading of the UE to a network device, either autonomously or as the result of a poll, the buffer occupancy measure may take account of the contents of the RLC buffer as well as the PDCP and application buffers.
Although the functions have been described in accordance with a three level protocol stack, the allocation of the functions described to a specific layer of position within the protocol is not intended to be limiting as it will be evident to a person skilled in the art that such functions may be moved between functional or protocol layers without departing from the teachings herein.
In another example, the scheduler 48 may differ from a round robin algorithm in order to effect a prioritization of the emptying of the queues, based on design criteria related to the quality of service to be allocated to each of the applications and associated data streams. Such an allocation may be based on multiple criteria, and may be dynamically altered. Examples of such criteria are: user perception of delay in service, criticality of data for safety, and the like.
A method for managing data flows in a wireless communications system may include providing a TCP/IP protocol to configure data from one or more applications into data flows, transferring data blocks from each active application program data stream to a corresponding data queues and maintaining the data queues, and a scheduling algorithm to transfer data from the data queues to an output buffer, in accordance with a data request from the output buffer.
In an example, shown in
In another example, shown in
A method 800 of computing the amount of data present in the plurality of data queues Qn is shown in
Schedulers may have algorithm attributes where the size M of the block of data removed from a queue Qk may depend on the priority established for the queue, or the queue may be serviced more than once per scheduler cycle or less than once per cycle. The priority may also relate to the amount of data remaining in the queue and in the application buffer.
Although this description has been approached from the viewpoint of the UE it will be appreciated that a similar approach may be adopted at the other end of a radio link for the return path, and such a return link may provide for connections to a number of UE, each UE having a number of active data steams.
The present invention has been explained by way of the examples described above; however, it should be understood to the ordinary skilled person in the art that the invention is not limited to the examples, but rather that various changes or modifications thereof are possible without departing from the spirit of the invention. Accordingly, the scope of the invention shall be determined only by the appended claims and their equivalents.
