1、 Rec. ITU-R M.1741 1 RECOMMENDATION ITU-R M.1741*Methodology for deriving performance objectives and its optimization for IP packet applications in the mobile-satellite service (Questions ITU-R 85/8, ITU-R 87/8, ITU-R 112/8 and ITU-R 233/8) (2006) Scope This Recommendation stipulates the methodology
2、 for deriving performance objectives and its optimization for IP packet applications in the mobile-satellite service. The guidelines for the performance parameters and objectives for physical and MAC layers, the methodology for derivation of the performance objectives, and the guidelines for the opt
3、imization of TCP performance in IP packet data applications in the mobile-satellite service are provided in Annexes 1, 2 and 3 to this Recommendation, respectively. The ITU Radiocommunication Assembly, considering a) that Internet Protocol (IP) packet transmission has become one of the major applica
4、tions in modern communication networks including mobile-satellite systems; b) that hypothetical reference circuits, technical characteristics, performance objectives and availability requirements have been stipulated for conventional mobile-satellite services (MSSs) in a number of existing Recommend
5、ations; c) that technical characteristics and performance should be defined on the basis of IP packet layers, in addition to basic digital transmission performance of the MSS bearer link; d) that Recommendation ITU-R M.1636 stipulates definitions for reference models and performance parameters as a
6、technical basis for the development of IP packet applications in the MSS; e) that the guideline for the performance objectives of the physical and link layers of MSS systems serving IP packets, together with the methodology to derive performance objectives for IP packet transmission for the system a
7、re required to facilitate effective use of spectrum resources by the system; f) that the transmission control protocol (TCP) is one of the most widely spread transport layer protocols over IP, where special attention is required for optimization of the operational parameters if it is applied to a sy
8、stem with large transmission delay such as MSS links, recommends 1 that the guidelines for performance objectives in Annex 1 should be used for IP packet transmission in the MSS; 2 that the methodology in Annex 2 should be applied for derivation of the performance objective for IP packet application
9、s in the MSS; 3 that the guidelines in Annex 3 should be taken into account when the operational parameters of TCP over IP packet transmission in the MSS are determined. *This Recommendation should be brought to the attention of Radiocommunication Study Group 4 and Telecommunication Standardization
10、Study Group 13. 2 Rec. ITU-R M.1741 Annex 1 Guidelines for performance objectives for IP packet transfer in the MSS 1 Introduction Terminals in the MSS have a number of performance requirements regarding employed protocols to ensure that the characteristics of the satellite link (such as large delay
11、, variable error rate, a periodic disruption) are tolerated and adequately handled. The protocols to be considered, which operate under IP packet transmission over the satellite link, include the physical (PHY) and the media access control (MAC) layers. This section identifies the characteristics an
12、d performance of PHY and MAC layers that can contribute to the performance of IP packet transmission over the MSS link. 2 Physical layer (Layer 1) 2.1 Packet-based multiple access channel Radio channels and their resources are generally accessed by a number of mobile-satellite terminals to connect t
13、o satellite access gateways through an MSS link for IP packet transmission. The sharing mechanisms of the radio channels and their resources for the packet-based multiple access involve a combination of techniques. One of the possible resource management approaches for the mobile-satellite terminals
14、 to access the radio channels is to utilize the channels in the forward direction on a time-division-multiplex (TDM) basis, and in the return direction on a time-division-multiple-access (TDMA) basis. In this case, each satellite access gateway manages a set of forward (to-mobile) and return (from-m
15、obile) channels. In addition, mobile-satellite terminals may support one or more transmit/receiver pairs. The TDMA for the physical layer requires certain transmission scheduling protocols, which are described in detail in 4.4. 2.2 Physical layer roles The PHY layer is responsible for the transfer o
16、f an information bit stream over the satellite link, and includes the following functions: At a transmitter: encoding, scrambling and interleaving; modulation of the encoded bit stream; transmission of the modulated signal on a multiplexed channel. At a receiver: reception of a modulated signal; dem
17、odulation of the signal into an encoded bit stream; decoding, descrambling and de-interleaving. The PHY layer transfers an entire frame together with the measured information concerning parameters of the satellite signal such as timing and power level. Rec. ITU-R M.1741 3 3 MAC layer (Layer 2) The M
18、AC layer is responsible for controlling access to the PHY layer (channel resources) by each connection and for the establishment and provision of the connection for the delivery of IP packets over the MSS link. The MAC layer generally performs the following functions: addressing of physical devices
19、or logical connections, and scheduling of resources for transfer of information between these entities in both the forward and return directions on a packet-based multiplexed channel; packing and unpacking of IP packets into MAC frames, including segmentation and reassembly, as required; buffering a
20、nd flow control of information from the upper layer (i.e. IP layer); automatic repeat request (ARQ) (if required for the particular, error sensitive connection). 4 Guideline for designing IP packet data transmission system in the MSS which may influence performance of the network layer (Layer 3) and
21、 higher This subsection describes the performance parameters and guidelines for system design of PHY and MAC layers for IP packet transmission in MSS, which can affect the performance of IP packet transmission over the MSS link. 4.1 Bit error ratio The bit error ratio (BER) for the MSS link can be d
22、erived from the design of the MSS link that depends on the physical layer characteristics of the link such as the modulation and the error coding scheme, transmitting power and link margin, receiver sensitivity and so forth. An example of BER performance for a 64 kbit/s bearer channel is shown in Ta
23、ble 1. TABLE 1 Example of BER performance Channel MSS terminal demodulator BER MSS gateway demodulator BER Measurement period 64 kbit/s 106 1061 500 s The values of the example in Table 1 are measured after synchronization of the signal has been established. All possible causes of the BER degradatio
24、n, such as phase noise, loss of synchronization lock, clock and cycle slips during the measurement are included but any degradation caused by burst error due to multipath fading, blockage, shadowing or adjacent channel interference is excluded. 4.2 Buffering and flow control The IP packets arriving
25、from the upper layer (i.e. IP networking layer) will be buffered at the MAC connection sub-layer before being transmitted by the packet-based multiple access channel of the PHY layer. The control mechanism and the size of the buffer will affect the performance of the IP packet transmission. For exam
26、ple, if the buffer size is sufficiently large, loss of packets due to the buffer overflow can be decreased, while the buffering delay can increase significantly if the buffer size is too large. Therefore, the mechanism of the buffer control, including the size of the buffer, needs to be determined c
27、arefully taking account of the trade-off between the loss and the delay of IP packet transmission. 4 Rec. ITU-R M.1741 4.3 ARQ The MAC layer offers a facility for reliable transmission, whereby information is delivered error-free and in-sequence, at the expense of an increase in packet transmission
28、delay. The mechanism for support of ARQ also needs to be selected such that the trade-off between error and delay performance can be well balanced. 4.4 Scheduling Scheduling is the process to allocate resources of the packet-based multiple access channel to each connection, which is managed by the s
29、atellite gateway. The simple method to allocate resources is first-in first-out (FIFO), in which priorities of all connections are considered as equal. In order to differentiate the priority of each connection, the scheduling process may be based on the quality of the service parameters of each conn
30、ection. In addition, the status of the queues for other connections might be used to determine the priority to transfer data in the scheduling process. Annex 2 Methodology for deriving IP packet transfer performance parameters in the MSS 1 Introduction The generic definitions of the performance para
31、meters and their objectives are provided in ITU-T Recommendations Y.1540 and Y.1541, respectively, for IP packet-based applications. For IP packet transmission using the MSS, dominant components for such performance parameters are contributed by the MSS section of the end-to-end communication path.
32、Taking account of the properties and the performance parameters of the MSS links defined in the previous section, this section summarizes some possible methodologies to derive the performance parameters of IP packet transmission over the MSS links. In this section, the MSS link or satellite section
33、refers to the satellite link between the satellite access terminals at both mobile and gateway earth stations via a satellite as depicted in Fig. 1 and Fig. 2 of Recommendation ITU-R M.1636. 2 IP packet error ratio The IP packet error ratio (IPER) is defined, in ITU-T Recommendation Y.1540, as the r
34、atio of total errored packets to the total successful and errored packets. The errored packet is defined such that the binary contents of the delivered packet information field do not conform exactly to those of the originated packet or that one or more of the header field(s) of the delivered packet
35、(s) is (are) corrupted. It would be possible to theoretically derive IPER caused by errored packets over the MSS link portion considering digital transmission performance parameters that are stipulated for MSS links. If the random error occurrence can be assumed, IPER, that is the probability that a
36、t least one bit will become errored for a packet, can be statistically derived as: 8_)1(1=sizePacketBERIPER (1) Rec. ITU-R M.1741 5 where: BER: bit error rate after applying the possible forward error correction (FEC) scheme Packet_size: size of IP packet in bytes transferred by the MSS link. For ex
37、ample, Recommendation ITU-R M.1476 stipulates that a performance objective for the MSS forming part of the ISDN defines that the BER should be less than 9 107. When an IP packet size of 1 500 bytes is assumed, the IPER for the system conforming to the ITU-R Recommendation becomes up to 1 102. It is
38、generally considered that there would be a trade-off relation between the reducing packet error and reducing transfer delay. Such a trade-off depends on the system design. For error-sensitive applications, ITU-T Recommendation Y.1541 stipulates that the performance objective for IPER is provisionall
39、y less than 1 104. To comply with the ITU-T Recommendation for the error-sensitive QoS classes, one possible approach is to apply an ARQ scheme to the MSS link to improve IPER at the expense of an increase in IP packet transfer delay. For delay-sensitive applications, no retransmission technique suc
40、h as ARQ can be applied. Therefore, IPER is determined only by performance of the satellite channel, that is, BER, and the size of each IP packet, Packet_size, that should carefully be determined through the system design. 3 IP packet loss ratio The IP packet loss ratio (IPLR) is defined, in ITU-T R
41、ecommendation Y.1540, as the ratio of total lost packets to the total transmitted packets. The causes for the loss of IP packets are, for example, misdirection of the IP packet due to the inconsistent update of routing tables, overflow of buffers at routers or packet transmission equipment, overload
42、 of routers, and so forth. Considering the MSS link section of the IP packet communication path, the primal cause, that is, the bottleneck for the IP packet loss would be the overflow of packets from the buffer at the interface of the satellite transmission equipment to connect the MSS link to the t
43、errestrial section. The ratio of IP packet loss due to the overflow of packets from the transmission buffer is dependent on the following parameters: IP packet arrival process; size of transmission buffer at an interface to connect the MSS link to the terrestrial section; scheme for IP packet schedu
44、ling over the MSS link. It is widely recognized that the arrival process of the IP packet data application is bursty in nature. For example, a traffic model is defined for web browsing for the purpose of evaluating radio transmission technologies in mobile telecommunications applications1. The model
45、 employs the so-called on-off packet arrival model. The on- and the off-periods correspond to a series of packets for a file download and the reading time of the file, respectively. The model is a good representation of the web browsing activities; however no analytical approach has been reported no
46、r established so far to evaluate the performance of IP packet data applications under the complicated packet data traffic model with the bursty nature. One possible approach to evaluate the IPLR under the bursty traffic is to apply the queuing simulation with appropriate assumption of the above para
47、meters. 1UMTS TR 101 112, “Selection procedures for the choice of radio transmission technologies of the UMTS,” April 1998. 6 Rec. ITU-R M.1741 Assuming the Poisson process for IP packet arrival and finite buffer with a FIFO scheduler, another possible approximation for the ratio of packet loss, IPL
48、R due to buffer overflow can theoretically be derived by the analysis of the M/M/1/K queue as follows: =KnnKIPLR0(2) where: K: buffer size in terms of the number of packets : traffic intensity, that is = h : packet arrival rate (number of packets per second) h: average holding time for packet transm
49、ission. 4 IP packet transfer delay (IPTD) IP packet transfer delay (IPTD) is the total transmission delay for an end-to-end IP connection. For IPTD, it is necessary for the end-to-end connection to be allocated properly to all sections that form the end-to-end connection. For the MSS link section, IPTDsatis defined in Recommendation ITU-R M.1636 as follows: bufferprocessingnnpropagationNnsatTTTIPTD +=+=,11(3) where: N: number of retransmissions by ARQ Tn,propagation: propagation delay of an MSS link for the n-th time transmission T
