Description
Method for transmitting messages, mobile station, base station and communications system
Introduction
The invention relates to a method for transmitting messages, a mobile station, a base station and a communications system.
In Document TDOC GP-032025 [2] the minimum features required in order to support MBMS service in GERAN are analysed. In Section 2 of [2] it is outlined that do nload-and-play as well as media streaming are the services that will generate the most MBMS traffic. While the codecs of the most streaming services can accept residual error rates in the region of 1%, MBMS services for download-and-play should operate error- free. An advantage of this 'type of background service is that the service can be delivered at bit rates different than the content rate. In a common Point-to-Point (P-t-P) GPRS trans- mission the acknowledged mode bearer would be used to transmit background and file data, as neither retransmission delay nor real-time playout is critical for download-and-play applications such as MMS .
MBMS services are transmitted in a broadcast or multicast mode to several users simultaneously over a common downlink channel using a Point- o-Multipoint (P-t-M) connection. Each terminal receives data transmitted on the same frequency and in the same time slot in the downlink. Using common downlink resources to distribute identical content to several users increases the system efficiency enormous the more receiving terminals are using the service simultaneously, in contrast to separate P-t-P connection.
State-of-the-art
Several Forward Error Correction (FEC) proposals [3], [4], [5] have been presented at GERAN #16 which introduce an outer coding scheme based on Reed-Solomon codes on the RLC layer as well as on the BM-SC layer to achieve reasonable throughput at SDU error rates about 10"2 to 10"3 for MBMS. Other proposals introduce repetition redundancy [6], [7], [8] or incremental redundancy [9], but do not perform as well as the Reed-Solomon schemes, or require significanti changes on the air interf ce. However, for background data transmission none of the proposals provides sufficient reliability. Therefore, an acknowledged mode MBMS bearer, comparable to the acknowledged mode in GPRS P-t-P transmission should be introduced [2], as only feedback and retransmission strategies can provide sufficient reliability. The goal is to exploit feedback information from the terminals to the BS and to reduce the residual error rate for the background services or at least to adapt somehow to the receiving conditions to the users in the cell.
1. P-t-P Retransmission
In [2] it was already proposed that for download services in MBMS with necessary residual error rate of 0%, two steps to transmit one file to several users in a cell are necessary. Firstly, a P-t-M connection using the common transmission channel in the downlink to all terminals is established and the data is transmitted in best-effort manner, possibly using one of the proposed outer coding systems to reach a residual block error rate of about 1% at a target channel-to- interference ratio (C/I) .
Secondly, after closing the P-t-M session all terminals which have detected lost packets establish P-t-P connections in acknowledged mode and request all lost packets individually on dedicated uplink and downlink channels for each terminal. All lost packets are retransmitted in separate P-t-P connections
to the terminals. This scenario is illustrated in Fehler! Verweisquelle konnte nicht gefunden werden..
Obviously, this strategy is a straightforward approach, as P- t-P connections are already supported in GPRS and P-t-P re- 1 transmissions provide in general low efficiency for distributing identical content to several users in broadcast and multicast systems, due to the fact that identical packets to be transmitted to N terminals will require N times more radio resources. As shown in Fehler! Verweisquelle konnte nicht gefunden werden. packet number 3 has to be transmitted twice, as both terminals could not receive this packet. In this example the channel needs to be accessed four times (four separate transmissions are required) to retransmit all users' lost packets. Obviously, all sequence numbers of lost packets have to be stored at the receiver until the end of the P-t-M transmission, what will require appropriate storage at the terminals, as the P-t-P retransmission session is likely to be performed after the P-t-M session.
2. P-t-M Retransmission
As outlined in Section 1 P-t-P retransmissions are not very efficient as packets with identical sequence numbers are retransmitted multiple times . Especially for scenarios with many users, e.g. in a football stadium, this transmission strategy turns out to be inefficient or even infeasible. For GERAN let us define a channel use as the transmission of one RLC/MAC radio block using one frequency, which results in the occupation of four radio bursts. Assuming the sender is aware of which packets are lost at each receiver, the basic idea of P-t-M retransmissions is these packets are also transmitted over a common physical channel (all terminals listen to the same frequency and time slot) . Retransmissions to multiple receiver sites using a common channel were already proposed in the early 80ies for reliable satellite file transfer and were shown to increase the overall system performance [10] .
The benefit of such systems results from the fact that a specific packet which was requested by several terminals will require only one channel access for its retransmission, rather than N channel accesses assuming N terminals requesting this packet. The system efficiency will increase and channel resources which are left over can for example be assigned to voice users or to other services. Fehler! Verweisquelle konnte nicht gefunden werden. shows the principle of the scenario. In the illustrated case, packet number 3 was requested for retransmission by both terminals. Actually, the channel uses can be reduced in comparison to the P-t-P scenario depicted in Fehler! Verweisquelle konnte nicht gefunden werden. as packet 3 is only transmitted once. In general, the number of channel uses for P-t-M retransmissions is therefore less or equal to the P-t-P retransmission scenario. Equality is given if disjoint packets are requested by the terminals, which is not expected to happen in the general case. Furthermore, one major advantage of the P-t-M retransmission scenario in contrast to the P-t-P scenario is that the retrans- missions can be performed immediately after an unsuccessful packet transmission, as the transmission bearer for transmission and retransmission is identical. This reduces the memory requirement in the receiver, as sequence numbers of packets do not have to be stored until the end of the file transmis- sion.
3. P-t-M Retransmissions exploiting incremental redundancy
In [11] it was shown that the performance of P-t-M retrans- mission systems can be significantly improved by exploiting incremental redundancy. Instead of retransmitting the lost packet itself, a redundancy packet is transmitted. A similar approach is used for hybrid ARQ protocols in point-to-point connections. In the P-t-M scenario the transmission of incre- mental redundancy, in contrast to simple packet repetition, reduces the number of required channel uses further.
Fehler! Verweisquelle konnte nicht gefunden werden. illustrates the benefits of the system. As each receiver lost two packets, each terminal needs only 2 additional redundancy packets to recover the packet losses due to the properties of the Reed-Solomon codes. Hence, only two channel accesses are required to recover the message for each terminal. In [11] an outer coding with punctured systematic Reed-Solomon codes is proposed to recover from packet losses in a P-t-M environ- ment .
This approach seems to be very promising for MBMS reliable data download. Furthermore, it is compatible with the already proposed outer Reed-Solomon coding. This system is even more attractive in case of many MBMS users: Whereas for the proposal in section 2 almost each packet has to be retransmitted, for P-t-M retransmission with incremental redundancy the performance (in terms of the overall number of required retransmissions) is almost identical to a single P-t-P retrans- mission to the worst user in the cell.
4. Feedback in P-t-M Systems
Although it is known as outlined in subsections 2 and 3 that feedback in p-t-m systems can significantly enhance the sys- tern efficiency, the application of such systems is very limited. Feedback in p-t-m system such as satellite or terrestrial broadcast systems is usually extremely complex or not possible, as the physical resources are usually not available in broadcast systems, the access protocols are rather complex for a high number of users, and, the handling of uplink messages by broadcast device is usually infeasible due to the high number of users . Some system approaches have been proposed which use a dedicated system for the uplink. For example broadcast systems are paired with telecommunication systems, such that the up-
link is realized by a data line within the telecommunication system. However, the complexity of these approaches hinders or at least limits the deployment of these approaches. To reduce the complexity especially in case of many users, ap- proaches have been discussed which only allow transmitting a only a subset of users in the uplink. In addition, the transmitter may control the feedback transmission by polling selected users. However, this again increases the complexity of the transmitter significantly, as users have to be identified in the system and appropriate polling policies have to be applied. More details on feedback in broadcast systems are discussed in Section 3.
Invention In the sections above retransmission schemes for multicast and broadcast scenarios have been reviewed. The advantages of P-t-M retransmissions, especially using incremental redundancy have been briefly outlined. For more details on the excellent performance in different scenarios we also refer to [11], [12] and [13]. However, for all cases in Section 2 as well as for the systems discussed in [10], [11], [12], [13] perfect knowledge on the status of each packet for each user in terms of positive (ACK) and negative acknowledgements (NAK) at the sender is assumed, i.e. a dedicated feedback channel which carries ACK and NAK messages from each user terminal to the sender is considered. This feedback channel could be considered as a P-t-P feedback channel, which carries feedback messages from a single receiver to the sender, or a shared channel where dedicated resources for a single receiver might be available on a TDM or CDM basis. Moreover, in the general case it is supposed that the feedback channel is able to carry not only the ACK and NAK messages, but also additional information which specifies the sequence number and/or the terminal the feedback message belongs. The draw- backs and requirements of state-of-the-art broadcast and multicast systems can be summarized as follows:
a dedicated feedback channel for each user must be available, each packet must be acknowledged (positively and/or negatively) , - feedback channel must carry ACK and NAK messages, a sequence number of each packet is transmitted to the sender, and, possibly an identification of the requesting terminal has to be transferred. However, a major problem of traditional broadcast and multicast systems, like TV broadcast or satellite broadcast, is the missing feedback channel. Usually it is assumed that a different system can provide a low-rate feedback, but in general retransmission strategies for P-t-M systems are not very widespread, due to the complexity of the feedback channel infrastructure and the handling of many feedback messages at the sender .
It is an object of the invention to transmit messages to a plurality of mobile stations in an effective way.
This object is solved by the features of the independent claims .
Advantageous embodiments are defined by the features of the dependent claims .
For P-t-M retransmission modes in MBMS in GERAN, the support of individual uplink channels for each terminal is too com- plex for most application scenarios especially in case of the supports of many users. However, in contrast to satellite and terrestrial broadcast systems, uplink and downlink transmission resources in communication systems such as GERAN or UTRAN are always paired. In MBMS, the corresponding uplink resources are not used. Therefore, we propose an approach to utilize these resources simultaneously for all mobile terminals to transmit feedback information. We introduce a Common
Uplink Feedback Channel (CUFCH) . Basically, the investigated approach relies on sending not-acknowledgement (NAK) messages from users who could not receive the recent message. Users having received the information do not send acknowledgement (ACK) messages.
All terminals use the same resources such as frequency, time slot, and/or code (e.g. in a CDMA system) in the uplink to inform the BS about the lost packet and send NAKs simultaneously. Actually, for the BS it is in general not of interest which terminals lost the corresponding packet, especially if a retransmission for each lost message has to be performed anyways in the p-t-m system. Basically, feedbacks are sent by all mobile terminals as uplink bursts on the CUFCH within explicitly assigned resources, where the assignment of re- sources depends on the sequence number of the corresponding downlink message. If a mobile terminal does not receive a downlink message, it will send an uplink burst within the corresponding uplink resources. Otherwise, nothing is transmitted. If an uplink burst is detected within the assigned resources at the base station, it is assumed that the corresponding message transmitted on the downlink was lost for at least one mobile terminal . If more than on terminal sends uplink bursts within the same resources, a superposition of uplink burst will occur. In any case, the BS listens if some power or energy is received in a certain timeslot. If the received power or energy exceeds a certain threshold, it is assumed that there are users in the serving area, who request a retransmission of some recent information. If the power is below a certain threshold, it is assumed that the message has been received correctly by all or most users. There would be further methods which can estimate the power or energy, and/or signal level and/or number of Not-Acknowledged- Messages and/or a combination thereof. All these information can be used to decide whether the (most) users are satisfied and whether a retransmission should be conducted.
Sending one NAK message for each downlink message is still extremely costly in terms of radio resources, especially in
systems such as GERAN, where the mobile terminals usually operate in simplex mode, i.e. they are not capable to transmit and receive messages at the same time. Even worse, most of the time some guard interval has to be accepted for switching between sending period and receiving period.
For incremental redundancy systems as presented in subsection 3, not for each message in the downlink an uplink message has to be returned, but only for a set of messages. The base station can react as follows to these uplink mes- sages: Ignore the messages and just continue sending new data. Repeat the lost messages. Use this message for adaptation of code parameters for future code words according to the current user topology which can be estimated from this feedback messages. Send redundancy symbols, if a certain receive power threshold in the uplink channel is exceeded. Note that the lower the power threshold is, the higher the reliability of the service. - A combination of all aspects.
The timing and frequency of these feedback messages on the uplink with respect to the transmitted code word sequence on the downlink depends on the number of messages in the set of messages. In addition, it can be dependent on the application requirement of the actual MBMS service.
A specific realization for the organization of downlink and uplink messages would be to have an arrangement whereby in the frames where the mobile terminal can transmit a NAK, no MBMS data is transmitted in the downlink.
A specific realization for the uplink burst would be use only part of a complete radio access burst. For example the middle part, a random part for each user, or any other part of the complete uplink burst might be used to send an uplink mes- sage .
For P-t-M systems with incremental redundancy in the downlink the position of the information symbols and redundancy part within a code word has to be signalled. The redundancy part can be a redundancy symbol of a Reed-Solomon code word, a part of incremental redundancy for RCPC codes, or any other redundancy symbol obtained from puncturing a channel code. The position of information and redundancy parts has to be signalled. The signalling can either be done implicitly by using some transmission order, external sequence number, etc., or explicitly by indicating the position explicitly in the packet header. In addition, the sequence number of the code word to which this information or redundancy portion is assigned to, has to be signalled. This can also be done implicitly by applying some transmission order or explicitly by specifying the sequence number of the code word.
Benefits of the Invention or of embodiments of the invention
The benefits of the invention or embodiments of the invention can be summarized as follows: - the introduction of an uplink channel allows to introduce reliable data services in p-t-m scenarios such MBMS over GERAN the introduction of a common uplink channel allows to send feedback messages even in case of a high number of participating users. The combination of the common uplink channel with incremental redundancy allows to increase the efficiency in the system significantly, as the system can provide reliable services and the throughput of the system is al- most identical to the throughput in p-t-p transmission of the worst supported user. The uplink messages can be used as an add-on, as the back channel messages can basically be ignored, if the reliability of the service is not guaranteed. In addi- tion, the feedback information can be used to get a basic idea on the user topology in the serving area. The
channel coding rate and the throughput can be adjusted appropriately .
Prior art and the invention are described by way of example with reference to the attached drawings wherein:
Figure 1 shows P-t-M packet transmission to multiple users,
Figure 2 shows Retransmission of lost packets to individual terminals using a P-t-P connection,
Figure 3 shows P-t-M retransmission of lost data packets using a common physical channel,
Figure 4 shows P-t-M retransmissions exploiting incremental redundancy, Figure 5 shows Principle idea: Incremental redundancy in broadcast systems.
Example of the system
The basic idea of sending incremental redundancy for multiple users is shown as an example in Fehler! Verweisquelle konnte nicht gefunden werden.. Thereby, message 1, message 2, and message 3 form a set of messages. The broadcasting of all messages to user 1 and user 2 yields in the successful transmission of message 1 to both users, the successful transmission of message 2 to user 2 and the successful transmission of message 3 to user 1. The transmission of message 3 to user 1 as well as the transmission of message 3 to user 2 ails. Then, in case of p-t-m retransmission according to subsection 2.2, user 1 would send a NAK for message 3 and request immediate retransmission of the same message. The same would hap- pen for message 2 of user 2. However, for the incremental redundancy case, user 1 and user 2 would send only one a NAK for the set of messages after message 3 on the common uplink channel. The transmitter detects the NAK and forms, in this
case, a simple parity check packet of the last three messages. As this parity message is received successfully for both users, both users can reconstruct their missing message. Applying more sophisticated codes than parity check allows recovering more than one packet loss .
References
[I] 3GPP TR 25.992-140, "Multimedia Broadcast/Multicast Ser- vice (MBMS); UTRAN/GERAN requirements".
[2] TSG GERAN Tdoc GP-032025, "Minimum requirements for MBMS support in GERAN" Ericsson, TSG GERAN #16, New York (USA), 25-29 August 2003.
[3] TSG GERAN Tdoc GP-031994, "Outer coding for MBMS", Sie- mens, TSG GERAN #16, New York (USA), 25-29 August 2003.
[4] TSG GERAN Tdoc GP-032100, "Outer Reed-Solomon Coding on RLC Layer for MBMS over GERAN", Siemens, TSG GERAN #16, New York (USA), 25-29 August 2003.
[5] TSG GERAN Tdoc GP-032101, "Outer coding on RLC layer for MBMS over GERAN: Extension to multislot mode", Siemens, TSG GERAN #16, New York (USA), 25-29 August 2003. rδ] TSG GERAN Tdoc GMBMS-0300002, "Performance of MBMS Radio Bearers", Nokia, TSG GERAN MBMS Workshop, Espoo (Finland), 12-13 May 2003. [7] TSG GERAN Tdoc GMBMS-0300007, "On MBMS bearer definition", Siemens, TSG GERAN MBMS Workshop, Espoo (Finland), 12-13 May 2003.
[8] TSG GERAN Tdoc GMBMS-0300014, "Bit rate and retransmission aspects for P-t-M MBMS in GERAN", Ericsson, TSG GERAN MBMS Workshop, Espoo (Finland), 12-13 May 2003.
[9] TSG GERAN T-doc GMBMS-030008, "New channel coding schemes for MBMS", Siemens, TSG GERAN MBMS Workshop, Espoo (Finland), 12-13 May 2003.
[10] S.B. Calo and M.C. Easton, "A broadcast protocol for file transfer to multiple sites", IEEE Trans. Commun. Vol. COM-29, pp. 1701-1707, Nov. 1981.
[II] J.J. Metzner, "An improved broadcast retransmission protocol", IEEE Trans. Commun. Vol. COM-32, pp. 679-683, Jun. 1984. [12] K. Sakakibara and M. Kasahara, "A multicast hybrid ARQ scheme using MDS codes and GMD decoding", IEEE Trans. Commun. Vol. 45, no. 12, pp. 2933-2940, Dec. 1995.
[13] R.H. Deng, "Hybrid ARQ schemes for point-to-multipoint communication over non-stationary broadcast channels", IEEE Trans. Commun. Vol. 41, no. 9, pp. 1379-1387, Sep. 1995.