A method in a communication system
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a method, a system and a station in a system using a contention based multiple access protocol. The method relates to controlling a transmission time a transmitting station may transmit data to a receiving station over a medium.
BACKGROUND AND RELATED ART
A contention based multiple access protocol is a type of protocol that as opposed to so- called conflict free protocols, does not guarantee that a transmission is successful. Conflicts may occur, e.g. collisions of transmissions from several users, and need to be resolved by the protocol. Furthermore these protocols do not require a centralised control station and therefore they are comprised in a group of protocols called distributed protocols. Contention based protocols are however only a subset of all distributed protocols. (Under the term distributed protocols some other protocols would also fit such as for example token ring protocol. The invention that will be disclosed in this application could also be applied for token ring protocols.)
Examples of contention based multiple access protocols are ALOHA, Carrier Sense Multiple Access (CSMA), CSMA with Collision Avoidance (CSMA/CA), which is used for IEEE 802.11 WLANs, and CSMA with Collision Detection (CSMA/CD). The CSMA protocols are characterized by always sensing the medium, i.e. listening to the medium and determining if the medium is free or occupied, before a transmission is performed. Contention based multiple access protocols are described and defined in the book "Multiple Access Protocols" by R. Rom and M. Sidi, Springer- Verlag, 1990, ISBN 0-387-97253-6.
IEEE802.il systems employ a Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA)-based MAC (Multiple Access Control). A comprehensive description of the IEEE 802.11 MAC is provided in the 802.11 standard (ANSI, IEEE Std 802.11, 1999 Edition, "Local and Metropolitan Area Networks: Wireless LAN"). In short, a station that wishes to transmit a packet first senses the medium. If the medium is sensed idle for a certain minimum time, a so-called Distributed Inter Frame Space (DIFS) time, the packet is transmitted. If the medium is busy, a backoff timer is set to a random number. The backoff timer is then decreased when the medium is sensed idle, whereas whenever the medium is sensed busy, a deferral state is entered, and the backoff timer is not decreased. When the backoff timer expires, the packet is transmitted. If the packet is successfully received, the receiver responds with an acknowledgement to the transmitter. The acknowledgement is sent a Short Inter Frame Space (SIFS) after the data frame is received. Since SIFS < DIFS, no other user will access the medium during this time. If no acknowledgement is received, either because the packet itself or the acknowledgement was lost, the transmitter generates a new backoff, and retransmits the packet when the backoff timer has expired. Even if the packet was successfully acknowledged, the transmitter must generate a backoff and wait for it to expire before transmitting the next packet. To avoid congestion, when collisions occur, backoff values are drawn from distributions with larger and larger expectations for eveiy retransmission attempt. For the « h transmission attempt, the backoff time is drawn from the uniform distribution U[0,min(32*2""1 - 1, 1023)]. The backoff time is measured in units of slot times, which for 802,1 lb are 20μs long. The procedure described above is exemplified in a flowchart in Figure 1.
The article "Performance Anomaly of 802, 1 lb" by Martin Heusse et al. published in IEEE Infocom 2003 (Full reference: Heusse, ; Rousseau, F.; Berger-Sabbatel, G.; Duda, ; INFOCOM 2003. Twenty-Second Annual Joint Conference of the IEEE Computer and Communications Societies. IEEE , Volume: 2 , 30 March - 3 April 2003 Page(s): 836 -843) analyzes the performance of the IEEE 802.11b wireless local area networks. It is disclosed that when some mobile hosts use a lower bit rate than the others, the performance of all hosts is considerably degraded. The reason is that the
basic CSMA/CA channel access method guarantees an equal long term channel access probability to all hosts, and low bit rate users will capture the channel for a majority of time since users with lower data rate transmit for longer time. This penalizes other hosts that use a higher bit rate and the aggregate throughput is reduced.
It is known, and utilized e.g. for WCDMA evolved (or HSDPA), that the aggregate throughput can be increased if users are scheduled when then channel qualities are good. This is often referred to as multi-user diversity. The drawback is that the delay may be increased for users with poor radio conditions. Not all services tolerate such delay variations.
In WO 03/049485 a scheduling algorithm that controls the usage of channel based on channel quality is described. Suggested channel quality measures are among others; channel rate and Frame Error Rate (FER) that are available in 802.11. In WCDMA evolved, the access point or Node B is aware of the radio link quality of each mobile station, either explicitly through channel quality reports, or implicitly from power levels used for other channels. The Node B may then schedule the user that has the best, or relatively best, channel quality. There is no central scheduler in 802.11. Even though WO 03/049485 addresses WLAN it does not describe how it shall be applied on 802.11, Multi-user diversity appears to be an attractive solution for 802.11, at least for services with somewhat relaxed delay requirements. It is however not obvious how to introduce it. There are no standardized radio quality measurement reports as for WCDMA evolved.
TERMINOLOGY
The following terminology is used in this application: the term station refers to an entity capable of receiving and/or transmitting data, i.e. including some transmitter and/or receiver means. An access point is a station that provides access to services of other networks, e.g. a fixed local area network or the Internet. Alternative terms for the access point include base station or Node B. A teraiinal, or mobile terminal, is a
station used by end-users of the communications system. Alternative terms for the terminal include mobile, mobile station, or user equipment. The invention is applicable but not limited to the case where the communications system is an IEEE 802.11
WLAN.
SUMMARY
An object of the invention is to provide a method and a station in a system using a contention based multiple access protocol, e.g. ALOHA, Carrier Sense Multiple Access (CSMA), CSMA with Collision Avoidance (CSMA/C A) or CSMA with Collision Detection (CSMA/CD), providing the system with an increased aggregate throughput compared to conventionally used systems.
This is achieved in a method according to claim 1, a station according to claim 12, a system according to claim 25 and a computer program according to claim 26 and 27.
Because of the increased expected waiting time per transmitted bit given to stations transmitting data with a comparatively low transmission rate these transmitting stations will be given a decreased transmission time compared to stations transmitting data with higher transmission rates and the aggregate throughput of the system will be increased.
The object is also achieved in a method according to claim 10 and a station according to claim 23. Hereby the transmission time for low rate users is decreased since they can be forced to another access.
In one embodiment of the invention the expected waiting time per transmitted bit is controlled by adjusting the random access wait times assigned to the transmitting stations, i.e. adjusting the time before access attempt after the medium is detected as free. One possibility is to vary the backoff time or the DIFS time in dependence of the ttansmitting station's transmission rate.
In another embodiment the controlling of expected waiting time per transmitted bit is instead performed by segmenting data into smaller packets depending on the station's transmission rate.
Other preferred embodiments are described in the dependent claims.
BRIEF DESCRIPTION OF THE DRAWINGS
Fig. 1 is a flowchart showing the backoff and deferral procedure in a contention based system.
Fig. 2 shows a communications system with terminals and access points according to the invention.
Fig. 3 is a flowchart of the different method steps according to a first embodiment of the invention.
DETAILED DESCRIPTION OF THE EMBODIMENTS
According to the invention an expected waiting time per transmitted bit assigned to transmitting stations is controlled such that stations transmitting data with a comparatively low transmission rate are given an increased expected waiting time per transmitted bit in relation to stations transmitting data with a higher transmission rate. Stations could for example be mobile terminals, other user terminals or access points in a system using a contention based multiple access protocol, e.g. ALOHA, Carrier Sense Multiple Access (CSMA), CSMA with Collision Avoidance (CSMA/CA) or CSMA with Collision Detection (CSMA/CD).
This could be achieved by providing longer random access wait times to stations having comparatively low transmission rates and shorter random access wait times to
stations having comparatively high transmission rates. Random access wait time refers to a total time the transmitting station has to wait before transmission.
In a first embodiment this is achieved by providing a back off time distribution that is a function of the transmission rate of the data to be transmitted. It is also possible to provide at least two different fixed back off time distributions and draw a back off time from the one or the other of the distributions depending on the data transmission rate. By providing stations having low transmission rates with a higher expected back off time, they less likely will access the medium. Hereby the low rate stations will not occupy as big portion of the transmission medium as if all stations draw back off times from the same distribution. In this application we use the term medium or transmission medium to refer to a generic communications resource shared by the stations. Typically this is a limited frequency band, but it may also be further divided in time and in code dimensions.
In a second embodiment of the invention the expected waiting time per transmitted bit is controlled by providing different Distributed Interframe Space, DIFS, times to the data to be transmitted depending on the transmission rate of the data. By providing data having a high transmission rate with a short DIFS time this data does not have to wait as long as data having longer DIFS times to be transmitted. Hereby high rate data is prioritised and the medium will not be dominated by data with low transmission rates. This embodiment could also be combined with the first embodiment, i.e. both back off times and DIFS times can be varied depending on transmission rate.
In a third embodiment the expected waiting time per transmitted bit is controlled by segmenting data into smaller packets for stations transmitting data having a comparatively low transmission rate. In this embodiment all packets are provided with random access wait times from the same distribution, i.e. all packets have the same probability to access the medium. However low rate packets are instead segmented into smaller packets and will therefore not occupy the medium as much as they would if all packets' sizes were equal.
Also according to the invention the transmission time is controlled by forcing low rate stations to transmit over another access. This is possible only in case of a multi access system. Such multi access systems are for example described in M. Frodigh et al., "Future Generation Wireless Networks", in IEEE Personal Communications, pp. 10- 17, Vol. 8, Issue 5, October 2001.
A detailed description of the embodiments with references to the drawings follows below.
A communication system according to the invention is shown in Fig. 2. Three terminals 1 are shown communicating with two Access Points 3 comprised in the communication system.
The terminals 1 comprise all the usual necessary communication equipment. Shown in Fig. 2 are only those parts that are necessary for performing the inventive steps of the method according to the ήivention. A transceiver means 7 and a rate selecting means 9 which is connected to the transceiver means 7 are shown in Figure 2. The transceiver means 7 retrieves data to be transmitted from the terminal 1 from a processing means in the terminal 1. The rate selecting means 9 selects a transmission rate for the data to be transmitted. This selection could be based on different factors. For example the transceiver means 7 can estimate the channel quality by measuring the signal strength of beacons sent from the Access Points 3 and the selection of a transmission rate can be based on this. A high channel quality allows a high transmission rate and a low channel quality requires a lower transmission rate. It could also be a randomly selected rate and the rate can be decreased from a previously used rate if it is detected in the terminal that the previous transmissions have been unsuccessful. If for example 802.1 lb is used in the system there are four available rates, i.e. 1, 2, 5.5, and 11Mbps. The supported subset of these rates, i.e. those rates being used by this access point, are transmitted from the access points in beacons and the terminals have to select one of the rates for the transmissions,
According to the invention the terminal 1 furthermore comprises a medium usage controlling means 13 which is connected to the rate selecting means 9 and to the transceiver means 7. The medium usage controlling means 13 is adapted to control the transmission time on the medium assigned to the data to be transmitted from the terminal. The controlling is based on the transmission rate for this data and said rate is retrieved from the rate selecting means 9.
There are a few different embodiments of the medium usage controlling means 13 according to the invention. In a first, second and third embodiment the medium usage controlling means 13 controls an expected waiting time per transmitted bit of the data to be transmitted. In the first embodiment this is achieved by providing at least two different back off time distributions in the medium usage controlling means 13. A back off time is a time period that is randomly selected from a certain backoff time distribution each time data is to be transmitted and the medium is sensed busy, or when transmission follows immediately after a transmission from the same station. If the transmitting medium is occupied at the time for the transmission the backoff time is counted down before the next tiy to transmit the data. See also the description of backoff times in the background chapter. Data that is going to be transmitted with a comparatively low transmission rate should according to the invention draw a back off time from the back off time distribution having the longest back off times and data with a higher transmission rate should draw a back off time from a distribution with shorter backoff times. Hereby data with low transmission rates have to wait a longer time before it is transmitted and the overall usage of the medium by low rate transmissions is decreased.
Preferably the medium usage controlling means 13 according to the first embodiment . comprises a back off time distribution that is a f nction of the transmission rate that is retrieved from the rate selection means 9. For example, in the IEEE802.11 standard where a backoff time distribution according to fBO=U(0, min((CWmill+l)*2T-1-l,CWmax))
is used, where CWmin and CWmaxare predefined constants, 31 and 1023 respectively for 802.1 lb and T is the number of transmission attempts, CWmjn and/or CWmax could be made dependent on the rate. In 802.1 lb there are four different possible rates and accordingly CWmin and CWmax could according to the invention take four different values. It is also possible that two or more of the rates are given the same CWmιn and CWma - A simple example for IEEE802.1 lb could be to use CWm;n values of 31, 63, 127 and 255 for the transmission rates 11, 5.5, 2, and 1Mbps respectively and keep CWmax constant.
The 802.1 le standard already employs different values for CWmιn and CWmax for different service classes. According to the invention CWmιn and CWmax could be made dependent on both the service type and the transmission rate. For delay sensitive seivices the dependency on the rate should be smaller than for delay insensitive services. For example if one service is video and another service is e-mail, the video seivice should be prioritised and be given low backoff times since this is a more delay sensitive service than e-mailing, i.e. according to the invention the rate dependency of the backoff times need not to be used for all service classes.
In a second embodiment of the invention the control of an expected waiting time per transmitted bit is achieved by providing in the medium usage controlling means 13 different DIFS times to the data to be transmitted depending on which rate has been selected for this data. The DIFS time is as the backoff time a time period to wait before the transmission of data. The station will always wait the DIFS time before transmitting data while the backoff time is only used when the medium is occupied. See also the description of DIFS in the background chapter. According to the invention a shorter DIFS time is given to data with a high transmission rate and a longer DIFS time is given to data with lower transmission rates. Hereby the probability to enter the medium and transmit is lower for data with lower transmission rates.
The terminals comprise thus also means for sensing if the medium is free and means for counting down the back off and DIFS times. These means are connected to the transceiver means and in the first and second embodiment of the invention also to the medium usage controlling means.
For 802.11 the back-off time and DIFS time are together forming a total time period before access attempt after the medium is free. This time period is also called a random waiting time. The principles of the first two embodiments are applicable also on other contention based multiple accesses protocols with other mechanisms defining this random waiting time.
In a third embodiment of the invention the control of an expected waiting time per transmitted bit is achieved by providing in the medium usage controlling means 13 a packet segmenting means that is adapted to segment the packets of the data to be transmitted into smaller packets if the selected rate for this data is lower than a predefined threshold. All packets are provided with random access wait times from the same distribution, i.e all packets have the same probability to access the medium. By segmenting low rate packets into smaller packets the transmission medium will not be occupied by low rate transmissions to the same extent as when all packets have the same size and are provided with random access wait times from the same distribution. In this embodiment the medium usage controlling means 13 needs to be connected to the processing means in the terminal 1 for retrieving the data to be transmitted. The data is then after the segmenting procedure forwarded to the transceiver means 7 for transmission. With the same probability to access the medium the time allocated to use it can thus still be controlled by varying the packet size. One packet segmentation algorithm could be to equalize segment transmission time (channel occupation per packet) independent of modulation/channel rate: ^size- ^max -K-ch' iιax
Where Smax is the segement size used for the highest transmission rate, Rc is the actual channel rate and Rmax is the maximum channel rate, e.g. 11Mbps for IEEE 802.1 lb. For example; With 802.1 lb Rmax=l 1Mbps for which the largest segment Smax is used.
If the transmission rate is 5.5Mbps, Smax 12 segment is used resulting in an equal transmission time per segment as for 11Mbps transmission rate. Alternatively, a smaller set of segment sizes could be used, e.g. for 802.11b one smaller size for 1 and 2Mbps, and a larger size for 5.5 and 11Mbps.
This packet segmentation can be straight forward applied also on a token ring multiple access protocol. Each time the token is received, one packet with a size defined by the transmission rate is sent.
Load in the network, i.e. how many stations that are fransmitting at the same time, can be used as a trigger for activating the above described segmentation mechanism, i.e. only perform segmentation in high loaded situations. Load can in the same way be used as a trigger for activating the use of different backoff and DIFS times according to the first and second embodiments of the invention.
In a fourth embodiment of the invention the medium usage controlling means 13 is adapted to control a handover to another access depending on the transmission rate. This embodiment can only be applied in a multi access systems with multi access capable mobile stations. If the terminal has selected a low transmission rate the medium usage contiOlling means 13 can force the terminal to handover to another access. Preferably the load in the system should first be considered and a handover only be performed if the load is high. The handover could also be controlled from the network , i.e. if an access point finds out that a terminal transmits data at a low rate and the load in the network at the same time is high the access point can force this teraiinal to handover to another access. In this case the medium usage controlling means is positioned in the access point instead of in the terminal.
However all four of these described embodiments could be combined in any way to achieve the best result.
In a communication system as shown in Figure 2 data is transmitted both in the uplink and downlink. Above the uplink procedure was described. The downlink transmission, i.e. from access point to terminal is performed in almost exactly the same way. In Figure 2 two access points 3 according to the invention are shown. The access points 3 comprise essentially the same components as was described for the terminals and of course all other necessary equipment for an access point in a communication system. An AP transceiver means 21 is connected to an AP rate selecting means 23. An AP medium usage controlling means 25 is connected to both the AP rate selection means 23 and the AP transceiver means 21. The functions of these parts are the same as was described for the same parts in the terminal with essentially one exception. This is that the access point has to keep in mind and separate transmissions to different terminals. The AP rate selection means 23 can select a rate based on received channel quality and needs in this case to keep a check on which terminal the data is intended for and which was the last received channel quality from this terminal. As described above the rate selection could also be a randomly selected rate and/or just a rate that is adjusted to a lower rate when the transmission has failed a predefined number of times.
When a plural of terminals communicate with one AP it is known that the uplink gets a greater access probability than the downlink. Therefore, according to the invention, the back-off time, the DIFS time and the packet segmentation must not necessary be equal for downlink and uplink, In fact in certain situations the wait times and/or segment size can be adjusted to compensate for a difference in expected waiting time per transmitted bit between uplink in downlink. With 802.11, if one AP communicates with several terminals and the terminals are not communicating with each other the expected waiting time per transmitted bit will be higher for downlink than for uplink. By in the AP taking into account the number of temiinals fransmitted to, the back-off time, DIFS time and/or segment size can be adjusted to yield equal expected waiting time per transmitted bit for downlink and uplink. One way to balance the expected waiting time per bit between uplink and downlink is to manage one back-off process per terminal in the AP, i.e. while the backoff time is decremented for transmission to
one terminal, data could be sent to another terminal and the backoff times per terminal could be decremented simultaneously.
In Figure 3 a flow chart of the inventive method steps according to the first embodiment of the invention is shown. The steps are described in order below: In this example data is to be sent from the access point to a terminal in the system.
S31 : The AP transceiver means 21 retrieves data to be transmitted to one of the terminals in the system from a processing unit in the access point 3.
S33 : The AP rate selecting means 23 is informed about which teraiinal 1 the data should be sent to.
S35 : The AP rate selecting means 23 selects a transmission rate for the data to be sent. This selection could possibly be based on the channel quality of the last received data from this terminal. This channel quality is in this case retrieved from the AP transceiver means 21 and stored in the AP rate selection means 23.
S37: The AP medium usage controlling means 25 retrieves the selected rate from the AP rate selection means 23 and uses this rate to decide which backoff time distribution that should be used. Possibly the rate is inserted into the backoff time distribution formula being a function of the rate and the distribution would then automatically be different for different rates as described above.
S39: A backoff time is drawn from the distribution related to this transmission rate.
S41: If the medium is occupied when the data is to be sent from the AP transceiver means 21 the backoff time (and the DIFS time) is counted down before the AP transceiver means 21 tries to transmit the data again.
This was only a schematic overview of the method according to the first embodiment of the invention. As described above there are many different valiants.
The method according to the invention is implemented by means of a computer program product comprising the software code means for perfomiing the steps of the method. The computer program product is run in stations, e.g. teiminals and access points, in a system using a contention based multiple access protocol. The computer progiam is loaded directly or from a computer usable medium, such as a floppy disc, a CD, the Internet etc.
The present invention is not limited to the above-described preferred embodiments. Various alternatives, modifications and equivalents may be used. Therefore, the above embodiments should not be taken as limiting the scope of the invention, which is defined by the appending claims.