基于GEP变频和流量控制系统外文翻译

1、Efficient scheduling for PLC networks T. Chiras, P. Koutsakis and M. PaterakisPowerline communications (PLC) are currently being considered as an alternative for high-speed data communications and Internet access. Presented is an efficient bandwidth allocation scheme which significantly excels in co

2、mparison to the Extended Aloha Medium Access Control (MAC) protocol for the last mile access PLC networks.Introduction: The unparalleled growth of the Internet, combined with significant technological advance ments of VLSI and digital signal processing, and with the telecommunications market deregul

3、ation around the world, have made pow erline communications (PLC) a viable technology for next generation telecommunications. With multiple outlets in almost every room, everywhere, power lines are already the most pervasive network in the home or small office; therefore, they would be the preferred

4、 medium for providing broadband connection to rural or remote areas where telephone and cable connections may not exist. The market for PLC is twofold: to the home, or last mile access, and in the home, or last inch access 1.Relevant research on the MAC layer for PLC has focused more on in-home netw

5、orking 2, 3. The work presented in this Letter focuses on the last mile problem, and in troduces scheduling ideas which lead to significant improvements in network performance and user Quality of Service (QoS) compared to the Extended ALOHA 4, 5 protocol, for powerline communication networks.Propose

6、d scheduling scheme:Orthogonal frequency division modulation (OFDM) has been outlined as one of the best candidates for application in PLC systems with higher data rates, because of its excellent bandwidth efficiency 4 -6. We consider an OFDM transmission system which uses a number of subcarriers di

7、stributed in a frequency spectrum. The work presented in 4, 5 proposed three extensions of the basic ALOHA protocol in order to improve its performance on the PLC network: (a) piggybacking, which leads to a decrease in thesignalling delay ;thisisdefinedasthetimeneededfor the realisation of the reque

8、sting procedure for the transmission of a packet and includes the transmission of a request message to the base station and the reception of its response regarding the access rights; (b) use of data channe ls for signalling (from 7); (c) application of an adaptive backoff mechanism for use r access

9、to the signalling channel, as well as for user access to the data channels for signalling purposes. In our work, we also adopt the ideas of piggybacking and using data channels for signalling. However, we do not use the adaptive backoff mechanism proposed in 4, 5 for users to select the slot in whic

10、h they will transmit= retransmit their requests; instead, we propose three newideas, two regarding the slot selection mechanism and one regarding the channel selection mechanism for a PLC access network.A. Channel selection: We use and compare two mechanisms for channel selection in our study. The f

11、irst mechanism is similar to that used in 4, 5 and is named uniform channel selection in this Letter. With the use of this mechanism, each terminal which needs to access the medium selects uniformly one of the 15 channels (one for signalling and 14 for data transmissions); the only constraint is tha

12、t selection is made among channels which have at least one idle slot in the current channel frame (no transmission is scheduled in that slot from previous channel frame s). If the channel is congested, it is not taken under consideration in the channel selection process for the current frame.Our pro

13、posal for a second channel selection mechanism is named weighted channel selection . At the beginning of each channel frame the base station has full knowledge of the total number of idle slots in all the data channels and the signalling channel. Let this total number of idle slots be S . The probab

14、ility for a terminal to choose channel ,Y which has three idle slots in the current channel frame, in order to send its request is 3 =S. The respective probability for the signalling channel is equal to the total number of slots of the signalling channel (the slots of the signalling channel are by n

15、ature always idle at the beginning of a channel frame, as no information transmission takes place in them) divided byS . The weighted channel selection mechanism is designed in a way as to push requesting users to choose, in every channel frame, with greater probability the channels with the larger

16、number of idle slots, in order to decrease the probability of collisions in the system.B. Slot selection: After selecting a channel, a terminal needs to choose the slot in which it will transmit its request. We propose two different mechanisms for slot selection in our study. The first mechanism is

17、named uniform slot selection: after selecting a channel with M idle slots (this information is given to the terminals by the base station after the channel is selected), the terminal attempts to transmit in the first of these slots with a probability P ? 1 =M (for M ? 1, P is by default 0.5, otherwi

18、se a collision would be unavoidable). In the case of a successful transmission, a terminal acquires the specific slot for transmission in subsequent channel frames, while in the case of a collision the terminal continues to transmit in idle slots with the above-definedprobability. If the channel fra

19、me ends without the terminal having succeeded in its request transmission, the terminal repeats the processes of channel and slot selection for every new channel frame, for as long as it needs to gain access to medium. The second proposed mechanism for slot selection is named weighted slot selection

20、 and works as follows. After selecting a channel with M idle slots, the terminal creates the following group of M probabilities: 1=M ,1=M ,2= M ,3=M , .,( M 1) =M , and randomly associates each one of the idle channel slots with one of the probabilities in the group. If M ? 1, the probability is aga

21、in chosen by default to be equal to 50%. The weighted slot selection mechanism aims at offering the chance to requesting terminals to transmit their requests sooner, by using much higher transmission probabilities than the uniform slot selectionmechanism (at the cost of a possibly larger number of c

22、ollisions).With the use of the above ideas, four versions of our MAC protocol were examined: the uniform-weighted selection (U-W), referring to a uniform channel and weighted slot selection, the uniform-uniform selection (U-U), the weighted-uniform selection (W-U) and the weighted-weighted (W-W).Res

23、ults and discussion: The system parameters used in our work are taken from 4, 5 , in order to make a direct comparison with that work, which focused on data (Intern et) traffic. Since packet transmission in PLC should be made in very short frames so that the receiver can adapt to the rapid ( < 1

24、ms) ch anges in the PLC channel conditions, we chose to consider only packets with average size equal to 300 bytes and with mean interarrival time 0.96 s; this is the case defined in4, 5 as the frequent request case , and enables us to testour scheme under heavy traffic conditions. The offered traff

25、ic load per network station is 2.5 kbit =s. The packet sizes and interarrival times are geometrically distributed random variables. The frame duration is 47 ms, the slot duration is equal to 4 ms, the slot capacity is 32 bytes and the payload in each slot is 28 bytes. In our simulations we assume th

26、at a transmission channel offers a fixed data rate of 64 kbit =s, and that the network consists of 15 bidirectional transmission channels, one of which is reserved for signalling. We simulated one hour of network performance. Each simulation point is the result of an average of 10 independent runs (

27、Monte-Carlo method).rage packer sue 300 bytes50100150 2DG 250300350400M50500us哈鹏st X史00 善Fi誉.1 (_ orjif>£jf iscjn t)t frv .stin tertux /“tldtrvFig. 1 shows comparison of our results with the Extended ALOHA protocol of 4, 5. It is clear from the Figure that at low traffic loads the signalling

28、 delay achieved by all versions of our protocol is remarkably smaller than that achieved by the Extended ALOHA protocol. As the traffic load increases, signalling delay naturally increases also, due to the increase in the number of collisions in the network. Still, as shown in Fig. 1, signalling del

29、ay achieved by all versions of our protocol remains much smaller than that of the Extended ALOHA protocol, by several hundreds of ms. When comparing the results of the four versions of our protocols, W-U selection achieves the lowest signalling delay for low-to-medium traffic loads and U-W selection

30、 achieves the lowest signalling delay for medium-to-high traffic loads.As shown inFig. 2, for up to 150 users the use of the Extended ALOHA protocol provides almost identical performance in network utilisation with the four versions of our protocol. However, as traffic load increases, the network ut

31、ilisation achieved by the Extended ALOHA protocol is significantly smaller than those achieved by the four versions of our protocol, the difference between them exceeding 20% when the number of users ranges between 300 and 400 and remaining large even for higher traffic loads.匚 0rel£jH=壬 匚awe r

32、ae pack-et size 300 byte：一 W-U SQlocbon -e- LhU slecticn A UW WecbmW W selwiioni 5酊曲。ALQiHA1.0- 。.年 0.602 N50100150200250300350400450 5QCusersFig- 1 Cuinptif iyon W_加 廿 al hcrney in terms t)f fiefnurk uiiliMiiionSince the difference between the Extended ALOHA protocol and our protocol exists in the

33、transmission -retransmission algorithms used, it is clear that our proposed algorithms are the reason for which our schemes excel. More specifically, the adaptive backoff mechanism used in4, 5 has the inherent disadvantages that: (a) after the calculation of the retransmission interval, the terminal

34、 will attempt to retransmit in the newly calculated slot, disregarding any idle slots which may exist before the calculated one; on the contrary, in all versions of our scheme, a terminal which fails to transmit its request attempts to retransmit (with various probabilities) in each of the immediate

35、ly following idle slots, therefore our scheme achieves much better utilisation of the available bandwidth; (b) in 4, 5, after the end of a request procedure, a value of the collision counter (CC) used in the backoff algorithm is kept as a start value of CC for the next request procedure. Therefore,

36、once again, valuable slots are lost, as the terminal does not even attempt to exploit them. Our more' aggressive ' policy is the reasonfor the decrease in network utilisation for high traffic loads, which however does not affect our scheme ' s superiority, as shown in Figs 1 and 2. Caref

37、ul observation of the results presented in the Figures reveals that in all cases the W-U selection achieves the best results for low-to- medium traffic loads (number of users less than or equal to 250) and that the U-W selection achieves the best results for medium-to-high traffic loads (number of u

38、sers larger than 250). The reasons for this can be found in the inherent logic of each version of our protocol: (a) in the case of a high traffic load, idle slots are few; therefore, the probability with which the channel with the largest number of idle slots is chosen by requesting users is often q

39、uite high with use of the weighted channel selection , leading to an immediate increase of the collision probability in that channel. (b) the weighted slot selection offers to requesting terminals the chance to transmit their requests sooner, by using highertransmission probabilities; however, this

40、choice leads to a higher collision probability. Hence, in the case of low-to-medium traffic loads, where weighted channel selection is more effective as explained above, the weighted slot selection performs worse than the uniform slot selection, as the combination of the weighted mechanisms for both

41、 the channel and slot selection is shown by our results to be aoo' taggressive ' policy and to lead to inferior performance metric results. Based on the above, we conclude that the most efficient use of our protocol is a-mctwo one, in which WU selection is activated forlow traffic loads and

42、U-W selection is activated for high traffic loads. The implementation of this-mo dewoprotocol is very feasible, sincethe base station can easily make a rough estimation of the number of users in the system by multiplying the number of users currently transmitting in a frame with (1 =activity factor)

43、; the activity factor is 0.525 of the time, for each terminal. Even if the above estimation is not perfectly accurate (the utilisation of the signalling channel in the current frame should be taken into consideration for a more accurate estimation), it is still adequate as all versions of our protoc

44、ol have been shown from our results to be comparable in their efficiency, therefore even if the better of the two modes is activated with delay, this will have very small impact on the user QoS metrics.#The Institution of Engineering and Technology 2007 15 November 2006Electronics Letters online no:

45、 20073394 doi: 10.1049/el:20073394T. Chiras and M. Paterakis ( Department of Electronic and Computer Engineering, Technical University of Crete, Greece)P. Koutsakis ( DepartmentofElectricaland Computer Engineering, McMaster University, Canada) E-mail: polkece.mcmaster.caReferences1 Majumder, A., and

46、 Caffery, J. Jr.:' Power linenconamunicatioverview ' , IEEE Potentials, 2004, 23, (4), pp84-2 Lin, Y.-J., Latchman, H.A., Liu, J.C.L., and Newman, R.:' Periodiccontention-free multiple access for broadband multimedia powerline communication networks ' . Proc. 9th Int. Symp. on Power

47、Line Communications and its Applications (ISPLC), Vancouver, Canada, 20053 /4 Hrasnica, H., and Lehnert, R.:' Extended AlohaPondrHybridReservation MAC Protocols for broadband powerline communicationsaccess networks . Proc. IXthVWII orld Telecommunications Congress (WTC), Pa

48、ris, France, 20025 Hrasnica, H., Haidine, A., and Lehnert, R.: Broadband powerlinecommunications: network design (John Wiley & Sons, 2004)6 Dostert, K.: Powerline communications (Prentice Ha)ll PTR, 20017 Koutsakis, P., and Paterakis, M.: Highly e-fdfiactiaenint tveogircaetionover medium high ca

49、pacity wireless TDMA channels , Wirel. Netw., 2001, 7, (1), p. 43-54PLC 网络的高效调度T.芝拉士， P. koutsakis Paterakis 和 M.电力线通信（PLC ）目前正在考虑作为一个高速数据通信和互联网接入方案。提出了一种有效的带宽分配方案，明显优于在扩展ALOHA 介质访问控制（ MAC ）协议的比较“最后一公里”接入 PLC 网络。简介 :在世界各地,这是个无与伦比的互联网发展时代，他结合重大进步的技术VLSI 和数字信号处理事件以及电信市场的管制。他使战俘erline 通信 （PLC） 为下一代电信可行

50、的技术。许多中国的媒体几乎在每一个房间,到处都是, 电线已经是最普遍的网络了在家庭或小型办公室; 因此 ,他们将是首选的媒介来提供宽带连接到农村或偏远地区的电话和电缆，连接可能不存在。市场对PLC 是双重的:家里,或 “最后一英里”访问,在家里,或 “最后一英寸”访问 1 。相关研究对 PLC 的 MAC 层则更加注重家庭网络 2 、 3 。 这封信侧重于 “最后一英里” 问题 ,在定级调度思想，对电力线通信网络导致显著的改善网络性能和用户的服务质量（QoS）相比，扩展迎宾 （4 、 5）协议。a- 提出了调度方案:正交频分modula -优化 （OFDM） 列出了作为一个最好的候选人应用 P

51、LC 系统和更高的数据率,因为其具有优良的带宽效率4 - 6 。我们考虑一个OFDM 传输系统,它使用许多subcarriers 分布在一个频谱。提出的工作 4,5提出三个扩展基本的迎宾协议中 , 以改善其性能在 PLC 网络 :（一）盗用,导致下降;thisisdefinedasthetimeneededfor thesignalling 延迟的实现请求过程中传输一个数据包,包括传输请求消息到基站和接收的响应对于访问权限;（b）使用数据channe ls的信号（来自7）;（c） 的应用程序的一个自适应补偿机制来使用 r 访问信号通道 ,以及用户对数据的访问通道信号的目的。在我们的工作,我们也

52、采用的思想和使用数据盗用通道信号。然而 ,我们不使用自适应补偿提出了机制 4,5 为用户选择槽中他们将传输=转播他们的请求 ; 相反 ,我们提出三个新的想法,两个关于槽选择机制和一个关于信道选择机制为PLC 接入网。a .通道选择: 我们使用和比较两个机制通道选择在我们的研究中。第一个机制是类似于 4,5, 被命名为统一的通道选择在这封信。使用这种机制,每个终端需要访问介质选择一致的15 个频道（一个用于数据传输的信和14）; 惟一的限制是在渠道选择 ,至少有一个空闲插槽在当前信道帧（没有传播计划,从先前的通道帧年代插槽） 。如果这个频道是拥挤的 ,它是没有被考虑在渠道选择过程为当前帧。我们的

53、建议第二信道选择机制是命名加权通道选择。 在每个通道的开始帧基站拥有完全的知识总数的空闲插槽在所有的数据通道和信号通道。让这空闲插槽的总数是年代。为终端的概率选择通道Y,它有三个空闲插槽在当前信道帧，为了发送其请求3 = S。 各自的概率为信号通道等于总数量的插槽的信号通道 （ 的槽信号通道是由大自然总是闲置开头的一个通道帧,因为没有信息传输发生在他们）byS 分裂。加权信道选择机制设计以 “推 ”要求用户选择,在每一个通道帧,以更大的概率渠道,并与大量的闲置槽以减少碰撞的概率在系统。b .槽的选择 :选择一个频道后,终端需要它将选择发送的请求槽中。我们提出两种不同的机制选择，在我们的研究中,

54、槽的第一个机制是命名统一槽的选择:在选择一个通道与M 空闲插槽 （此信息给终端通过基站的信道选择后 ）,终端试图传播在第一个插槽和一个概率P?1 = M（M?1,P 是默认情况下 0.5,否则一个碰撞将会是不可避免的 ） 。对于一个成功的传输、终端获得特定的插槽,在后续帧传输通道,而对于碰撞终端继续传播在空闲插槽与上述定义概率。如果通道帧结束没有终端已经成功在其请求传输,终端重复的过程通道和槽选择每一个新通道帧,只要它需要获得介质。第二个提议机制选择被命名为加权槽槽的选择和工作如下。在选择一个通道与M 空闲插槽,终端创建以下群M 概率 : 1 = M,1二,2 =,3 =,，（M 1）= M

55、随机的每一位同事空闲信道插槽和一个组中的概率。如果 M?1,概率是再次选择默认情况下等于50%加权槽选择机制旨在提供机会请求终端传输他们的请求早,通过使用更高的传输概率比均匀槽selectionmechanism（在可能的成本更多的碰撞）。使用上面的想法，有四种 MAC 协议检测：统一加权选择（ U-W ） ，它指的是统一通道和加权槽选择、统一的均匀选择（ U-U ） 、加权统一选择（ W-U ）和加权的加权选择（ W-W ） 。结果与讨论:系统参数用于我们的工作取自4,5,为了使直接比较,工作主要集中在数据 （实习生等）交通。因为包传输在PLC 应该会在很短的帧,以便接收方可以适应快速（ 1

56、 ms）ch anges 在 PLC 信道条件下 ,我们选择只考虑数据包的平均大小等于300个字节 ,意味着 interarrival 时间 0.96秒;这是如此定义在4,5 “频繁请求案件 ”, 使我们能够测试我们的方案在拥挤的交通条件。提供交通负载/网络站是2.5 kbit = s 。 包的大小和interarrival 时间几何分布的随机变量。 帧持续时间是47 ms,槽时间等于4 ms,槽容量是32个字节和负载在每个槽是28字节。在我们的模拟我们假设一个传输信道提供一个固定的数据速率为 64 kbit = 年代 ,和网络由 15 双向传输通道 ,其中之一是预留给信号。 我们模拟一个小时

57、的网络性能。 每个模拟点的结果平均10 独立运行（蒙特卡罗方法）。Fig. I C(frnptirisfin offiw schemes rn 如用工 切上曾行。门吟 delay图1显示了比较我们的结果与扩展的迎宾协议4,5。很明显从图可看到在低流量加载信号延迟通过所有版本的我们的协议是显著的小于通过扩展的迎宾协议的。随着交通荷载的增加，信号延迟自然增加，由于在网络中的碰撞的次数的增加。仍然，如图1所示，信号延迟的所有版本的协议的实现仍然比扩展ALOHA协议要小得多，短数百女士当比较结果四我们的协议版本，w-u选择达到最低的信号低到中等流量负载和u-w选择延迟达到最低的信号延迟中高流量负载。As ol.如图inFig2所示、150用户使用扩展的迎宾协议提供了几乎相同的性能在网络利用率和我们四个版本的协议。然而，随着交通负荷的增加，网络利