Wireless channel allocation in a base station processor

ES 2 227 196 T3 description

Wireless channel assignment in a base station processor.

Background of the invention

Wireless network infrastructure equipment is increasingly being used to enable computer devices to communicate via a wireless medium with a wired network such as the Internet. In a wireless data network, a plurality of local computing devices, such as personal computers, are supported by means of wireless subscriber access units. A subscriber access unit provides a wireless radio link to a base station processor. The base station processor is also connected to an Internet access portal that provides a connection to a wired network. Similar to a cellular telephone network, the base station processor allocates a plurality of wireless channels on a demand basis to provide transmission of messages to and from the subscriber units. Wireless channels are assigned to messages sent and received from the subscriber unit on behalf of the local computing device.

In a typical base station processor, wireless channels are a scarce resource that are shared by subscriber units. Messages are frequently queued, pending the availability of a channel. Additionally, wired networks typically employ techniques to detect the speed with which a receiver is processing messages. These techniques reduce congestion by avoiding overloading a receiver by reducing the rate at which messages are sent, thereby reducing performance. Such techniques can interpret the queuing of messages in the base station processor as congestion on the wired network, thereby reducing performance. In particular, the protocols used in the wired network do not lend themselves well to efficient communication over wireless connections.

In a TCP / IP network, for example, congestion control techniques such as slow start, congestion avoidance, fast retransmission, and fast recovery are employed. According to the slow start technique, as defined in Internet RFC 2581, an accept message (ack) is expected as a return message for each message sent. The number of bytes, or messages, sent is gradually increased as the acceptance messages are received in sequence. If acceptance messages are not received in sequence, additional messages will be sent less frequently, reducing performance. Queuing of messages in the base station processor is not, however, indicative of congestion in the base station processor. In contrast, queuing is indicative of the propagation delay inherent in wireless networks. This propagation delay is however interpreted as congestion by wired line protocols such as TCP / IP.

It would therefore be beneficial to provide a method and apparatus that can anticipate the arrival of the return message, and schedule a channel to be available to transmit the message through the base station processor so that performance on the wired network does not is reduced by the congestion control properties of the wired network protocol, such as slow startup.

Compendium of the invention

A system and method are provided for assigning wireless channels in a wireless communication system to support the transmission of messages between a subscriber and a base station processor. A latency period is determined corresponding to the rate of an expected return message from a responding node in response to an outgoing message sent from a sender by means of the base station processor. A latency manager in the base station processor calculates the latency period and stores the latency period in an allocation table. A programmer schedules a channel, prior to receipt of the return message, to be available at the end of the latency period indicated in the allocation table. At approximately the end of the latency period, the return message is received and the scheduler allocates a channel, as defined in the allocation table. The assigned channel is used to transmit the return message to or from the corresponding subscriber.

The latency manager calculates the latency period using a variety of transmission parameters defined in the wired line network protocol. For example, on a TCP / IP network, the transmission parameters used to calculate the latency period may include window size, window space available, average message size, number of pending accept messages, type of messages, number of messages received in the session, number of pending acceptance messages, maximum number of pending acceptance messages, and other transmission parameters.

Brief description of the drawings

The foregoing and other objects, properties, and advantages of the invention will become apparent from the more particular description that follows of preferred embodiments of the invention, as illustrated in the accompanying drawings, in which like reference characters refer to the same parts. throughout the different views. The drawings are not necessarily to scale, the emphasis being instead on illustrating the principles of the invention.

Fig. 1 is a block diagram of a communication system suitable for carrying out wireless channel assignment as defined herein;

Fig. 2 shows a base station processor in communication with a plurality of subscriber access units;

Fig. 3a shows the transmission of messages in the system of Fig. 2;

Fig. 3b shows a channel assignment table corresponding to the messages of Fig. 3a;

Fig. 4 shows a flow chart of channel assignment as defined herein;

Fig. 5a shows a web page search using the system of Fig. 1;

Fig. 5b shows the allocation table corresponding to the messages of Fig. 5a;

Fig. 5c shows a cadence diagram corresponding to the mapping table of Fig. 5b; Y

Fig. 6 shows a table of subscriber profiles for channel assignment as defined here. Detailed description of the invention

Fig. 1 is a block diagram of a communications system 10 operable for ac allocation

ES 2 227 196 T3 terminals in a wireless network as defined herein. The communication system includes a local computer device such as a personal computer 12, a subscriber access unit 14, a base station processor 16, and an Internet access portal 18. The personal computer 12 is in communication with the subscriber 14 via a wired connection 20. Subscriber 14 is in communication with a base station processor 16 via a wireless connection 26. The base station processor is in communication with an Internet access portal 18 via a wired link 24. The Internet access portal 18 is adapted for communication via a public access network such as the Internet.

The personal computer 12 can thus be provided with access to the network server 18, which can be any remote entity located on the Internet or another network, through a combination of the wired connections 20, 24 and wireless connections 26. wired connections 20, 24 are typically supported by a protocol such as TCP / IP or UDP. The wireless connection is supported by protocols such as the protocol described in US patent application Ser. pending entitled "Dynamic Frame Size Settings for Multichannel Transmission," published as PCT Application No. WO 99/44341, on September 2, 1999. Typically, personal computer 12 provides an Internet Protocol (IP) bundle to subscriber 14 via the wired connection 20, which can be for example an Ethernet type connection. Subscriber 14 unframes the IP packet and transfers the data in the IP packet to base station processor 16 via wireless connection 26 in accordance with a wireless handshake protocol. The base station processor 16 extracts the frames from the wireless connection and routes them, in IP batch form, through the wired line connection 24, to the Internet access portal 18. The subscriber 14 and the processor of the base station 16 are therefore considered as "end points" of the wireless connection 20.

Referring to Fig. 2, the base station processor 16 is shown in greater detail. The base station processor 16 is in communication with a plurality of subscribers 14a-14d. Additional subscriber units (14x) may be provided. Subscribers communicate with the base station processor via wireless channels 22a-22j shown. Additional channels 22 (x) can be added. As noted above, channels 22 are used to transmit messages to and from subscribers 14. A scheduler 28 allocates channels 22 on a demand basis, and allocates available channels to messages transmitted between subscribers 14 and the processor of the network. base station 16.

Channels 22 are unidirectional between subscribers 14 and switch 16, although multiple channels can be assigned to messages originating from or destined for a particular subscriber 14. In the example shown, channel 22a is assigned to transmit a message from the processor. from base station 16 to subscriber 14b. Channel 22b is assigned to receive a message in base station processor 16 from subscriber 14c, while channel 22c is assigned to send a message to subscriber 14c. Channels 22d and 22e are assigned to transmit a message to subscriber 14d, and channel 22f is assigned to receive a message from subscriber 14d. Typically, as noted above, scheduler 28 is rapidly assigning channels to subscribers to accommodate channel requests for messages to be sent to and received from subscribers 14.

Two dedicated channels, common to all subscribers 14, are used to initiate message traffic on one channel. A common access channel 30 is used by a subscriber 14 to request a channel from the base station processor 16. A common pager channel 32 is used to notify a subscriber 14 that a channel is being assigned. The messages are then directed by subscribers 14 to personal computer 12 or base station processor 16, depending on the address.

The base station processor 16 also includes a latency manager 34, for determining latency delays, and an allocation table 36, both of which are described below. In a typical message transmission, as noted above, a number of latency delays occur between the sender of the message and the receiver, or responding node. For example, a wireless propagation delay occurs when transmitting a message from the base station processor 16 to the subscriber 14 (Fig. 1). A network propagation delay occurs when a message is transmitted over the Internet or another publicly accessible network. Other latency delays are present, as will be described below. It is common in a protocol such as TCP / IP to expect a return message, typically an acceptance message, in response to a message sent to a responding node. In accordance with the invention as defined herein, the latency manager is embedded in the base station processor to calculate the latency delay and schedule a channel accordingly. Channel assignment refers to messages sent from a sender in any direction; the return message will be returned to the sender by the node that receives the message. Thus, a channel assignment for a return message will be predictably scheduled when a message is sent to or from subscribers 14.

Referring to Figs. 3a and 3b, a more detailed design of the base station processor 16 is shown, including the latency manager 34, the allocation table 36, and the scheduler 28. The latency manager 34 is a process that computes the latency delay associated with the return message sent by a responder mode 40. The allocation table 36 is a memory structure that stores an entry 38a, 38b for each channel allocation 22b, 22c, and associated latency times T₀ and T₀ + T_L. A scheduler 28 is a process that reads allocation table 36 and latency information to determine channel allocation to expected messages.

In a typical message transmission, personal computer 12 sends a connection request message to a responding node 40, as indicated by arrow 42. Message 42 is sent at time T0. Consequently, an entry 38a is written to allocation table 36 to allocate channel 22b with subscriber 14c at time T₀. As message 42 is received through channel 22b, latency manager 34 examines the message. The latency manager 34 determines that the type of the message is a TCP / IP connection request, and that therefore an acceptance message can be expected as a return message.

IS 2 227 196 T3

The latency manager 34 determines the latency period that will elapse before the reception of the return message in the processor of the base station 16. For example, the latency manager 34 determines that a delay ISP (Internet Service Provider) 44 will occur between the Internet access portal 18 and Internet 50, as indicated by AT₁; furthermore, a network propagation delay 46 will occur as message 42 is transmitted over the Internet 50, as indicated by AT₂; and then a delay 48 of the responding node will occur while the responding node 40 processes the messages and sends the return message, as indicated by AT₃. The latency period T_L 52 is therefore calculated by the latency manager, where T_L = AT₁ + AT₂ + AT₃. The latency manager then writes entry 38b to allocation table 36 to indicate that following the latency period, a return message 54 can be expected from responding node 40 to subscriber 14c.

Consequently, channel 22c is assigned at time T₀ + T_L for subscriber 14c. The return message 54 is sent by the responding node 40, and received by the base station processor 16, at time T0 + TL. In accordance with the allocation table 36, the scheduler 28 allocates channel 22c to transmit the return message 54 to the subscriber 14c.

In alternative embodiments, the channels are scheduled as a general grouping in the allocation table, and are not allocated to a specific subscriber until the return message is actually received.

In the example above, the latency manager 34 calculates the latency period TL 52 based on the type of the message and the corresponding expected return message. Many protocols, including the TCP / IP protocol, specify not only the return message, but also other transmission parameters. The way to determine the latency period therefore depends on a number of factors depending on the protocol in use. In a TCP / IP protocol, such factors may include transmission parameters such as window size, free space in the window, average message size, number of pending accept messages, message type, number of messages received in the session, number of pending acceptance messages, maximum number of pending acceptance messages, and other transmission parameters. For example, TCP / IP employs a sliding window performance enhancement property, as defined in Internet RFC 1323. Such properties can be used in conjunction with transmission parameters to enhance performance through a base station processor. as defined here.

A TCP / IP network can operate in accordance with the sliding window protocol in an effort to provide a reliable power supply while maximizing bandwidth. Under this protocol, both endpoints of a TCP / IP negotiate an acceptable window size. The window size designates a maximum number of octets that can be transmitted by a sending unit before receiving an acceptance from the receiving unit. Generally, the window is referred to in terms of the maximum number of batches not accepted. Once the sending unit receives an acceptance for the first batch in the window, it "slides" the window across and sends the next batch.

In message 42 sent in the example of Fig. 3a, the latency manager examines the TCP / IP batch non-destructively to determine the type of the message. Other aspects of the TCP / IP package, listed above, could also be examined to obtain transmission parameters, and use these parameters to determine the latency period 52 associated with the return message. In the examples that follow in Fig. 4 and 5a-5c, the latency manager 34 further includes a subscriber profile table 56 to store transmission parameters corresponding to each of the subscribers 14.

Referring to the flow chart depicted in Fig. 4 along with the system diagram of Fig. 3a, a message is received at the base station processor 16, as shown in step 100. The latency manager 34 examines the TCP / IP packet information, as described in step 102. A search is performed in the subscriber profile table to find the entry for the subscriber, as depicted in step 104. The corresponding transmission parameters are retrieved, as shown in step 106. The transmission parameters are updated to reflect the information from the TCP / IP packet examined in step 102, as described in step 108. A determination is made to indicate whether a return message is expected to complement the message, as represented in step 110. If no message is expected, the message is sent, as shown in step 120, and control returns to step 100 until the next message is received, as described in step 122. If a return message, the latency manager 34 calculates the latency period 52 using the subscriber's transmission parameters updated in step 108, as depicted in step 112. A new entry corresponding to the calculated latency period 52 is stored in the allocation table 36, as shown in step 114. The message is then sent to the responder node 40, as shown in step 116. Control returns to step 118 until the next message is received.

In Fig. 5a-5c another embodiment of the message transmission sequence of Fig. 3b is shown in greater detail. A connection request 42 is sent by personal computer 12 at time T₀. The latency manager 34 examines the batch information and determines the subscriber 14d. The latency manager looks up the transmission parameters of the subscriber 14d in the subscriber profile table 56, and updates the parameters accordingly to correspond to the new batch information. Latency manager 34 determines that a connection acceptance message 54 is expected as a return message. Latency manager calculates AT latency delays₁, AT₂ e AT₃as a result of the updated transmission parameters, and calculates the latency period T_TO as the result T_TO = AT₁ + AT₂ + AT₃. Latency manager 34 stores entry 58 in allocation table 36 to inform the programmer to allocate channel 22d for subscriber 14d at time TA, as indicated by entry 68 in cadence diagram 86.

The responder node 40 then sends the return message 54. The base station processor 16 receives the return message 54, and the

ES 2 227 196 T3 latency 34 examines the batch information. The latency manager looks up the transmission parameters of the subscriber 14d in the subscriber profile table 56, and updates the entry accordingly. The latency manager 34 determines that the type of the return message is a connection response acceptance, and that a request message will likely be sent from the personal computer as a return message.

As the message is sent to the subscriber 14d, the latency period is calculated as follows. The propagation time without AT wires₄ it is indicative of the latency associated with transmission via wireless connection 26 between base station processor 16 and subscriber 14d. The response time of the AT subscriber₅ is indicative of the latency associated with transmission via wired line 20 between subscriber 14d and personal computer 12. Consequently, latency manager 34 uses the subscriber profile table to calculate the latency period AT_Bfrom AT₄ + AT₅. A corresponding entry 60 is written to allocation table 36 to inform the programmer that it must allocate channel 22f for subscriber 14d at time T_B, as indicated by input 70 in cadence diagram 86.

The personal computer 12 sends an HTTP get message 78 after receiving the acceptance message 54. The corresponding transmission parameters are looked up in the subscriber profile table 56, and updated accordingly to correspond to the message 78. As a result of the updated transmission parameters, the latency manager 78 determines that an HTTP get accept message 80 and an HTTP data message 82 will likely be sent as return messages at the same time. Consequently, the latency manager calculates the latency period AT_C from AT_C = AT₁ + AT₂ + AT₃, and writes two entries to assignment table 36. Input 62 assigns channel 22d, and input 64 assigns channel 22e, for subscriber 14d at time TC, as shown by inputs 72 and 74 in the cadence diagram 86.

Upon receipt of the HTTP data message on switch 16, latency manager 34 determines that an HTTP data accept message 84 is the return message, and writes input 66 to assign channel 22f in T_D = AT₄ + AT₅, as shown by input 76 in cadence diagram 86.

An example of subscriber profile table 56 is shown in Fig. 6. Each input 86 is adapted to memorize transmission parameters 88 corresponding to messages received by a particular subscriber 14. Such parameters include window size, available window space, average message size, number of pending accept messages, message type, number of messages received in the session, number of pending acceptance messages, and maximum number of pending acceptance messages. Other parameters can be specified, such as defined in the TCP / IP protocol or other protocol, as used by the base station processor.

Those skilled in the art will readily appreciate that the programs defining the operations and methods defined herein can be supplied to the base station processor in many ways, including but not limited to: a) information permanently memorized in memory media that cannot be written, such as ROM devices, b) alterable information stored in memory media that can be written, such as floppy disks, magnetic tapes, compact discs, RAM devices, and other media magnetic and optical, or

c) information transferred to a computer via communication media, for example using baseband signaling or broadband signaling techniques, as in electronic networks such as the Internet or telephone modem lines. The operations and methods can be implemented in a piece of software executable from memory by a processor. Alternatively, the operations and methods may be performed in whole or in part using components of physical equipment (hardware), such as Application Specific Integrated Circuits (ASICs), state machines, controllers or other hardware components or devices, or a combination of hardware and software components.