88計(jì)算機(jī)專業(yè)應(yīng)用程序app設(shè)計(jì)相關(guān)有關(guān)外文文獻(xiàn)翻譯成品智能手機(jī)上的即時(shí)消息（im）應(yīng)用程序（app）的表征

1、<p>　　此文檔是畢業(yè)設(shè)計(jì)外文翻譯成品（ 含英文原文+中文翻譯），無(wú)需調(diào)整復(fù)雜的格式！下載之后直接可用，方便快捷！本文價(jià)格不貴,也就幾十塊錢(qián)！</p><p>　　外文標(biāo)題：Characterizing Instant Messaging Apps on Smart-phones</p><p>　　外文作者： Parth H. Pathak, Prasant Mohapat

2、ra</p><p>　　文獻(xiàn)出處:International Conference on Passive and Active Network Measurement, 2015: 83-95</p><p>　　英文2347單詞，12365字符，中文3895漢字。</p><p>　　Characterizing Instant Messaging Apps o

3、n Smart-phones</p><p>　　Parth H. Pathak, and Prasant Mohapatra</p><p>　　Abstract. Proliferation of smart devices has fueled the popularity of using mobile instant messaging (IM) apps at a rapid

4、pace. While the IM apps on smart-phones have become increasingly popular, there has only been a little research on understanding the characteristics of these apps. Because most of the IM apps use proprietary protocols, i

5、t is challenging to analyze their internal operations. In this work, we present a comprehensive characterization of mobile IM apps using experiments on LTE ce</p><p>　　1.Introduction</p><p>　　Re

6、cent years have witnessed a fast growing trend of using the new generation of mobile instant message (IM) applications such as Whats App, We Chat and Line on the smart-phones. Whats App, for example, is ranked as the thi

7、rd all- time-popular Android apps in Google’s Android app store [1] with a total of 590 million users in 193 different countries [2]. According to [3], the mobile IM apps have overtaken the Short Message Service (SMS) op

8、erated by cellular network carriers, with 19 billion messa</p><p>　　While the adoption of mobile IM apps are rapidly increasing, very little research has been done in characterizing them. This is because the

9、re are numerous challenges in characterizing the IM apps. First, compared to other types of mobile apps studied in [4–7], the IM apps involves much more user interaction such as typing, reading and user notification. Thi

10、s makes the automated characterization extremely difficult. The new set of features (e.g. typing and read notification) offered by the IM app</p><p>　　In this work, we present a comprehensive characterizatio

11、n of the popular IM apps for smart-phones using experiments on LTE cellular network. We address the challenges listed above by dissecting the operations of IM apps into many different states and then evaluate the energy

12、and the network efficiency of each of them. Some of the main insights provided by our study are as follows:</p><p>　　We find that sending and receiving typing notification are major contributors to the total

13、 energy consumption when the IM app is running in the foreground. Many IM apps use frequent periodic typing notification messages which result in very poor energy efficiency.</p><p>　　Today’s IM apps have ex

14、tremely low bandwidth efficiency (average amount of traffic per one character of user message). This is true even when the app is running in the foreground and has minimal requirement of maintaining the “online presence”

15、. This shows that while XMPP-like IM protocols offer efficient ways of maintaining “online presence”, the current IM apps show poor network efficiency when running in the foreground.</p><p>　　Because users s

16、pend significant amount of time on IM apps compared to other types of apps, simply switching to darker graphical interface can yield surprising energy benefits.</p><p>　　When the IM apps are running in backg

17、round, the method used to notify the user about incoming message has a significant impact on the energy consumption. </p><p>　　The rest of paper is organized as follows. We describe the experimental setup an

18、d the data collection in Section 2. The foreground and the background characterization results are presented in Section 3 and Section 4 respectively. Then we discuss the related works in Section 5. Section 6 concludes th

19、e paper.</p><p>　　2.Data Collection and Methodology</p><p>　　In this section, we first provide the details of data collection for different IM apps. We represent the operations of an IM app usin

20、g a state transition diagram. For each of the states, we will test 5 most popular IM apps, and pro?le the energy consumption and the network traffic generated.</p><p>　　2.1 State Transitions in IM App Usage&

21、lt;/p><p>　　As shown in Fig. 1, the operations of an IM app can be divided into 6 distinct states. When the users are in a conversation, the IM app runs in the foreground, occupying the entire screen. When the

22、users are using another app or when the screen is turned off, the IM app runs in the background but still keeps maintaining connections with its remote servers.</p><p>　　2.2 Methodology and Metrics</p>

23、<p>　　To get a comprehensive view of the characteristics of the selected IM apps, we test all the 6 states of the apps, by using a set of the most commonly used IM messaging literacy among college students [8]. In

24、 [8], the authors listed the taxonomy of the IM conversation topics. For example, the 5 most popular conversation topics are: emotional support, ?fictional people, video games, computers and shared interests. We picked o

25、ne conversation in each kind of the popular topics from the typical examp</p><p>　　The performance of the IM apps in the state with sending/receiving activities are mainly evaluated by two metrics: (i) Energ

26、y efficiency: energy consumption per character sent/received (Joule/character) and (ii) Bandwidth efficiency: the amount of network traffic generated per character sent/received (byte/character). In the idle listening st

27、ates, since there are no user intended messages, we will use the average energy consumption per hour (J) and the average network traffic per hour (K Byte) a</p><p>　　3In Conversation Sending/Receiving (ICS/I

28、CR)</p><p>　　We conducted a total of 12,600 runs of experiments by manually typing, and collected 2.4 GB of energy and network traffic traces. From the network traces, we observed that all the 5 selected IM

29、apps are built on the client/server architecture, where the message sender and the message receiver communicate indirectly through a certain number of servers. Although following the same architecture, the application la

30、yer protocols used by each app are quite different. By linking the server port number </p><p>　　As shown in Fig. 2, the ICS state can be divided into two phases: 1) typing the message & sending typing no

31、tification, and 2) sending the message & receiving the read notification.Correspondingly, the ICR state is also consisted of two phases: 1) receiving the typing notification, and 2) receiving the message & send

32、ing the read notification. Since the typing of a long message needs considerable amount of time, we can observe a time gap between the ?first and the second phase of the ICR state. </p><p>　　(a) Sending a Me

33、ssage with 120 Characters (b) Receiving a Message with 120 Characters</p><p>　　Fig. 2: Examples of Energy Traces of In Conversation States (Whats App) </p><p>　　Energy Characterization </

34、p><p>　　The energy consumption of “in conversation”in conversation states can be attributed to two factors: (1) the Graphical User Interface (GUI) and (2) user operations such as typing or sending messages etc.

35、</p><p>　　4The Background States</p><p>　　The performance of the IM apps running in the background is now characterized. We show the corresponding results in Fig. 6 - 8. Similar to the ICR state

36、, we can also observe the energy efficiency of the background receiving decreases if the the length of the messages decreases, as shown in Fig. 6. However, the reasons behind the phenomenons are quite different. In the b

37、ackground receiving states, we did not observe any typing notification nor read notification from the network traces. The main c</p><p>　　(a) BRON Energy Consumption Per Character (b) BROFF Energy Consumpt

38、ion Per Character</p><p>　　Fig. 6: The Energy Efficiency of Background Receiving</p><p>　　5Related Work</p><p>　　Traditional Messaging Services: There is a limited amount of prior w

39、ork on characterizing the performance of IM apps on smart-phones. PC-based IM apps (AIM and MSN) were characterized in [11] where authors studied network traffic related characteristics. Similarly, [12] characterized the

40、 users’ conversation styles of IM apps in workplace, by analyzing the SMS messages exchanged through AT&T’s cellular network. Note that different from both these efforts, we have attempted to characterize smart-p<

41、/p><p>　　App Profiling: There has been multiple research works on developing methods to profile smart-phone apps in general. This includes multi-layer pro?ling tool Pro?leDroid [7], Application Resource Optimiz

42、er (ARO) [13], energy measurement tool eprof presented in [14] and third-party API resource usage measurement tool API Extractor (APIX) presented in [15]. Different from these generic pro?ling tools, our focus in this wo

43、rk is to understand the network and energy characteristics specific to the IM a</p><p>　　Mobile IM Apps: Considering the research specific to mobile IM apps, [16] and [17] modeled user’s residence time on IM

44、 apps and typical message arrival rate. Based on these models, they derived energy consumption models of IM apps. The provided model, however, only provides a high-level coarse-grained behavioral analysis which is indepe

45、ndent of the operation of the underlying IM app. In this paper, our focus is on the operations of different IM apps. In other related work [18], the authors showe</p><p>　　6Conclusions</p><p>　　

46、By decomposing the operations of IM apps into 6 states, we characterized the energy and the bandwidth efficiency of IM apps. We also analyzed various operations of the IM apps, e.g. typing notification, read notification

47、, sending/receiving messages. Our analysis revealed there is still plenty of improvements necessary in the IM apps especially in the “in conversation” and the “background receiving”states to improve their energy and band

48、width efficiency. However, we observe that the background id</p><p>　　References</p><p>　　1. AppBrain, http://www.appbrain.com/stats/.</p><p>　　2. C. Clifford, “Top 10 apps for inst

49、ant messaging,” Entrepreneur, Dec 11, 2013.</p><p>　　3. BBC, http://www.bbc.com/news/business-22334338.</p><p>　　4. H. Falaki, D. Lymberopoulos, R. Mahajan, S. Kandula, and D. Estrin, “A first l

50、ook at traffic on smartphones,” ser. IMC ’2010, pp. 281–287.</p><p>　　5. S.-W. Lee, J.-S. Park, H.-S. Lee, and M.-S. Kim, “A study on smart-phone traffic analysis,” ser. APNOMS'2011, pp. 1 – 7.</p>

51、<p>　　6. Q. Xu, J. Erman, A. Gerber, Z. Mao, J. Pang, and S. Venkataraman, “Identifying diverse usage behaviors of smartphone apps,” ser. IMC’2011, pp. 329–344.</p><p>　　7. X. Wei, L. Gomez, I. Neamti

52、u, and M. Faloutsos, “Profiledroid: multi-layer profiling of android applications,” ser. Mobicom’2012.</p><p>　　8. C. Haas and P. Takayoshi, “Young people's everyday literacies: The language fea- tures o

53、f instant messaging,” Research in the Teaching of English, vol. 45, no. 4, pp. 378–404, May, 2011 .</p><p>　　9. I. A. N. A. (IANA), https://www.iana.org/assignments/.</p><p>　　10. VRMLSite, http

54、://www.vrmlsite.com.</p><p>　　11. Z. Xiao, L. Guo, and J. Tracey, “Understanding instant messaging traffic charac- teristics,” ser. ICDCS’2007, pp. 51–51.</p><p>　　12. E. Isaacs, A. Walendowski,

55、 S. Whittaker, D. J. Schiano, and C. Kamm, “The character, functions, and styles of instant messaging in the workplace,” ser. CSCW ’2002, pp. 11–20.</p><p>　　13. F. Qian, Z. Wang, A. Gerber, Z. Mao, S. Sen,

56、and O. Spatscheck, “Profiling resource usage for mobile applications: a cross-layer approach,” ser. MobiSys '11. ACM, 2011, pp. 321–334.</p><p>　　14. A. Pathak, Y. C. Hu, and M. Zhang, “Where is the ener

57、gy spent inside my app?: fine grained energy accounting on smartphones with eprof,” ser. EuroSys ’2012, pp. 29–42.</p><p>　　15. L. Zhang, C. Stover, A. Lins, C. Buckley, and P. Mohapatra, “Characterizing mob

58、ile open apis in smartphone apps,” in IFIP Networking Conference’ 2014, pp. 1–9.</p><p>　　16. Y. W. Chung, “Investigation of energy consumption of mobile station for instant messaging services,” ser. ISADS’2

59、011, pp. 343–346.</p><p>　　17. ——, “An improved energy saving scheme for instant messaging services,” ser. WiAd’2011, pp. 278–282.</p><p>　　18. E. J. Vergara, S. Andersson, and S. Nadjm-Tehrani,

60、 “When mice consume like elephants: Instant messaging applications,” ser. e-Energy ’2014, pp. 97–107.</p><p>　　智能手機(jī)上的即時(shí)消息應(yīng)用程序（App）的表征</p><p>　　Parth H. Pathak, and Prasant Mohapatra</p>&

61、lt;p>　　摘要：智能設(shè)備的普及快速地推動(dòng)了移動(dòng)即時(shí)消息（IM）應(yīng)用程序的流行。雖然智能手機(jī)上的即時(shí)消息應(yīng)用程序越來(lái)越受歡迎，但關(guān)于了解這些應(yīng)用程序特點(diǎn)的研究卻不是很多。由于大多數(shù)IM應(yīng)用程序使用專有協(xié)議，因此分析其內(nèi)部的運(yùn)作情況具有挑戰(zhàn)性。在這項(xiàng)工作中，我們運(yùn)用LTE蜂窩網(wǎng)絡(luò)上的實(shí)驗(yàn)，呈現(xiàn)移動(dòng)IM應(yīng)用的全面特征。我們將IM應(yīng)用程序的操作分解為多個(gè)獨(dú)立狀態(tài)，以方便我們能夠系統(tǒng)地研究它們。我們描述了每個(gè)狀態(tài)的能耗和帶寬效率并提供了

62、許多見(jiàn)解。我們的分析結(jié)果表明，IM應(yīng)用程序的信息輸入功能是耗能的主要方面。我們還發(fā)現(xiàn)與電子郵件和網(wǎng)上沖浪等其他應(yīng)用相比，當(dāng)前IM應(yīng)用的帶寬效率非常低。此次工作中提供的其他調(diào)查結(jié)果有助于提高IM應(yīng)用程序的耗能表現(xiàn)及其網(wǎng)絡(luò)性能。</p><p><b>　　1引言</b></p><p>　　近年來(lái)，在智能手機(jī)上使用Whats App（瓦次普），We Chat（微信）和L

63、ine（連我）等新一代移動(dòng)即時(shí)消息（IM）應(yīng)用程序的趨勢(shì)迅速增長(zhǎng)。例如，Whats App（瓦次普）被評(píng)為谷歌安卓應(yīng)用程序商店中第三個(gè)歷史上最受歡迎的安卓應(yīng)用程序[1]，在193個(gè)不同的國(guó)家/地區(qū)共有5.9億用戶[2]。 根據(jù)[3]，移動(dòng)IM應(yīng)用程序取代了由蜂窩網(wǎng)絡(luò)運(yùn)營(yíng)商運(yùn)營(yíng)的短消息服務(wù)（SMS），每天發(fā)送190億條消息，而SMS信息卻只有17.6億條。</p><p>　　雖然使用移動(dòng)IM應(yīng)用程序的用戶數(shù)量正在

64、迅速增加，但對(duì)于它們表征方面的研究卻很少。 這是因?yàn)樵诒碚鱅M應(yīng)用程序時(shí)存在許多挑戰(zhàn)。 首先，與[4-7]中研究的其他類型的移動(dòng)應(yīng)用程序相比，IM應(yīng)用程序涉及更多用戶交互，例如打字、閱讀和用戶信息通知。 這使得對(duì)這些應(yīng)用進(jìn)行自動(dòng)表征非常困難。與傳統(tǒng)的SMS服務(wù)相比，IM應(yīng)用程序提供的新功能（例如，鍵入和讀取通知）要復(fù)雜得多。此外，這些流行的IM應(yīng)用程序使用的應(yīng)用程序?qū)訁f(xié)議缺乏透明度。大多數(shù)當(dāng)前的IM應(yīng)用程序要么應(yīng)用自己的協(xié)議，要么修改現(xiàn)

65、有的標(biāo)準(zhǔn)（如XMPP）來(lái)進(jìn)行定制。 這使得理解應(yīng)用程序的基礎(chǔ)操作變得更加困難。</p><p>　　在這項(xiàng)工作中，我們使用LTE蜂窩網(wǎng)絡(luò)上的實(shí)驗(yàn)，對(duì)智能手機(jī)上的流行IM應(yīng)用程序進(jìn)行了全面的描述。我們通過(guò)將IM應(yīng)用程序的操作分解為許多不同的狀態(tài)，然后評(píng)估每個(gè)狀態(tài)的能耗和網(wǎng)絡(luò)效率，以解決上面列出的挑戰(zhàn)。我們的研究提供了以下的一些見(jiàn)解：</p><p>　　我們發(fā)現(xiàn)，當(dāng)IM應(yīng)用程序在前臺(tái)運(yùn)行時(shí)，

66、發(fā)送、接收、輸入信息通知是能耗的主要方面。許多IM應(yīng)用程序使用頻繁的輸入通知信息，這導(dǎo)致非常差的能耗效率。</p><p>　　現(xiàn)如今的IM應(yīng)用程序具有極低的帶寬效率（也就是用戶消息中每個(gè)字符的平均流量）。即使應(yīng)用程序在前臺(tái)運(yùn)行并且維持“在線狀態(tài)”的要求最低，情況也是如此。這表明雖然類似XMPP的IM協(xié)議提供了維持“在線狀態(tài)”的有效方式，但是當(dāng)前的IM應(yīng)用程序在前臺(tái)運(yùn)行時(shí)顯示出較差的網(wǎng)絡(luò)效率。</p>

67、<p>　　與其他類型的應(yīng)用程序相比，由于用戶在IM應(yīng)用程序上花費(fèi)了大量時(shí)間，因此只需切換到較暗的圖形界面即可產(chǎn)生巨大的耗能效益。</p><p>　　當(dāng)IM應(yīng)用程序在后臺(tái)運(yùn)行時(shí)，用于通知用戶有關(guān)消息的傳入會(huì)對(duì)能耗產(chǎn)生重大影響。</p><p>　　本文中剩余部分的結(jié)構(gòu)如下：我們?cè)诘?節(jié)中描述了實(shí)驗(yàn)裝置和數(shù)據(jù)收集。前景和背景表征結(jié)果分別在第3節(jié)和第4節(jié)中給出。 然后我們將在第

68、5節(jié)討論相關(guān)的工作。第6節(jié)總結(jié)本文。</p><p><b>　　2.數(shù)據(jù)收集和方法</b></p><p>　　在本節(jié)中，我們首先提供不同IM應(yīng)用程序的數(shù)據(jù)收集的詳細(xì)信息。我們使用狀態(tài)轉(zhuǎn)換圖表示IM應(yīng)用程序的操作。對(duì)于每個(gè)狀態(tài)，我們將測(cè)試5個(gè)最流行的IM應(yīng)用程序，并提供能耗和生成的網(wǎng)絡(luò)流量。</p><p>　　2.1 IM應(yīng)用程序使用狀態(tài)轉(zhuǎn)

69、換</p><p>　　如圖1所示，IM應(yīng)用程序的操作可以分為6種不同的狀態(tài)。 當(dāng)用戶進(jìn)行對(duì)話時(shí)，IM應(yīng)用程序在前臺(tái)運(yùn)行，占據(jù)整個(gè)屏幕。 當(dāng)用戶使用其他應(yīng)用程序或關(guān)閉屏幕時(shí)，IM應(yīng)用程序在后臺(tái)運(yùn)行，但仍保持與其遠(yuǎn)程服務(wù)器的連接。</p><p><b>　　2.2方法和指標(biāo)</b></p><p>　　為了全面了解所選IM應(yīng)用程序的特性，我們通

70、過(guò)在大學(xué)生中使用一組最常用的IM消息傳遞來(lái)測(cè)試應(yīng)用程序的所有6種狀態(tài)[8]。在[8]中，作者列出了IM對(duì)話主題的分類。例如，5個(gè)最受歡迎的對(duì)話主題是：情感宣泄、虛擬人物、視頻游戲、計(jì)算機(jī)和共享興趣。我們從[8]中總結(jié)的典型例子中選擇了每種流行主題中的一個(gè)對(duì)話，并創(chuàng)建了一個(gè)包含70條消息的數(shù)據(jù)庫(kù)。消息的長(zhǎng)度從4個(gè)字符到多達(dá)125個(gè)字符不等，其中字符可以包括字母，標(biāo)點(diǎn)符號(hào)和元話語(yǔ)標(biāo)記。為了減少隨機(jī)性的影響，每次運(yùn)行實(shí)驗(yàn)中每個(gè)消息的類型重復(fù)2

71、0次以計(jì)算平均值。我們?yōu)閮蓚€(gè)不同的用戶重復(fù)實(shí)驗(yàn)，以消除任何用戶特定的字體輸入特征。</p><p>　　具有發(fā)送/接收活動(dòng)狀態(tài)下的IM應(yīng)用程序的性能主要通過(guò)兩個(gè)指標(biāo)來(lái)評(píng)估：（i）能耗效率：發(fā)送/接收的每個(gè)字符的能量消耗（焦耳/字符）和（ii）帶寬效率：網(wǎng)絡(luò)量，每個(gè)字符發(fā)送/接收產(chǎn)生的流量（字節(jié)/字符）。在空閑偵聽(tīng)狀態(tài)中，由于沒(méi)有用戶預(yù)期的消息，我們將使用每小時(shí)的平均能耗（J）和每小時(shí)的平均網(wǎng)絡(luò)流量（K字節(jié)）作為評(píng)

72、估指標(biāo)。</p><p>　　3會(huì)話發(fā)送/接收（ICS / ICR）</p><p>　　我們通過(guò)手動(dòng)輸入共進(jìn)行了12,600次實(shí)驗(yàn)，并收集了2.4 GB的能量和網(wǎng)絡(luò)流量。 從網(wǎng)絡(luò)跟蹤中，我們觀察到所有5個(gè)選定的IM應(yīng)用程序都構(gòu)建在客戶端/服務(wù)器體系結(jié)構(gòu)上，其中消息發(fā)送者和消息接收者通過(guò)一定數(shù)量的服務(wù)器間接通信。雖然遵循相同的架構(gòu)，但每個(gè)應(yīng)用程序使用的應(yīng)用程序?qū)訁f(xié)議卻截然不同。 通過(guò)將服務(wù)

73、器端口號(hào)與互聯(lián)網(wǎng)號(hào)碼分配機(jī)構(gòu)（IANA）[9]的注冊(cè)表相關(guān)聯(lián)，我們發(fā)現(xiàn)We Chat，Whats App，F(xiàn)B Messenger，Line和Viber分別使用complex-main，XMPP，HTTPS，SSL和Virtual 現(xiàn)實(shí)建模語(yǔ)言（VRML）[10]。</p><p>　　如圖2所示，ICS的狀態(tài)可分為兩個(gè)階段：1）鍵入消息并發(fā)送打好的通知，2）發(fā)送消息并接收讀取通知。相應(yīng)地，ICR狀態(tài)也包括兩個(gè)階

74、段 ：1）接收打好的通知，以及2）接收消息并發(fā)送讀取通知。 由于長(zhǎng)消息的輸入需要相當(dāng)長(zhǎng)的時(shí)間，因此我們可以觀察到ICR狀態(tài)的第一階段和第二階段之間的時(shí)間間隔。 在時(shí)間間隔期間，無(wú)線電將被調(diào)整到尋呼信道（PCH）狀態(tài)以節(jié)省能耗。</p><p>　　（a）發(fā)送120個(gè)字符的消息 （b）接收120個(gè)字符的消息</p><p>　　圖2：會(huì)話狀態(tài)中的能耗追蹤示例（Whats Ap

75、p）</p><p><b>　　能耗表征</b></p><p>　　“會(huì)話中”狀態(tài)的能量消耗可歸因于兩個(gè)因素：（1）圖形用戶界面（GUI）和（2）用戶操作，例如鍵入信息或發(fā)送消息等。</p><p><b>　　4后臺(tái)狀態(tài)</b></p><p>　　在本節(jié)中，我們現(xiàn)在對(duì)在后臺(tái)運(yùn)行的IM應(yīng)用程序

76、的性能進(jìn)行表征。 我們?cè)趫D6-8中顯示了相應(yīng)的結(jié)果。與ICR狀態(tài)類似，我們還可以觀察到，如果消息的長(zhǎng)度減小，后臺(tái)接收的能量效率會(huì)降低，如圖6所示。但是，這些現(xiàn)象背后的原因是完全不同的。在后臺(tái)接收狀態(tài)中，我們沒(méi)有觀察到任何打印的通知，也沒(méi)有從網(wǎng)絡(luò)追蹤中讀取通知。能效下降的主因是通過(guò)一些方法通知用戶的開(kāi)銷，例如， 橫幅大小的通知、彈出窗口和圖標(biāo)標(biāo)簽。在BRON狀態(tài)下，Viber使用彈出窗口，而在BROFF狀態(tài)下，Line和Viber都使用彈

77、出窗口。彈出窗口導(dǎo)致這兩個(gè)應(yīng)用程序的能耗顯著增加，如圖6所示。</p><p>　?。╝）每個(gè)字符的BRON能耗 （b）每個(gè)字符的BROFF能耗</p><p>　　圖6：后臺(tái)接收的能量效率</p><p><b>　　5相關(guān)工作</b></p><p>　　傳統(tǒng)消息傳遞服務(wù)：在以前的研究中，關(guān)

78、于在智能手機(jī)上表征IM應(yīng)用程序的性能方面的很少。 基于PC的IM應(yīng)用程序（AIM和MSN）在[11]中進(jìn)行了表征，其中作者研究了網(wǎng)絡(luò)流量相關(guān)的特征。 同樣，[12]通過(guò)分析AT＆T的蜂窩網(wǎng)絡(luò)交換的SMS消息來(lái)表征用戶在工作場(chǎng)所的IM應(yīng)用的會(huì)話風(fēng)格。 需要注意的是與這兩次嘗試不同的是我們?cè)噲D描述智能手機(jī)IM應(yīng)用程序的特征，這些應(yīng)用程序徹底改變了人們?cè)诋?dāng)今時(shí)代的連接方式。</p><p>　　應(yīng)用程序分析：在此之前，

79、已有學(xué)者開(kāi)展了多項(xiàng)關(guān)于開(kāi)發(fā)智能手機(jī)應(yīng)用程序概述方法的研究工作。 這包括多層專業(yè)工具ProfileDroid [7]、應(yīng)用資源優(yōu)化器（ARO）[13]、[14]中提出的能耗測(cè)量工具eprof和[15]中提出的第三方API資源使用測(cè)量工具API Extractor（APIX）。與這些通用的專業(yè)工具不同，我們?cè)谶@項(xiàng)工作中的重點(diǎn)是了解IM應(yīng)用程序特有的網(wǎng)絡(luò)和能耗特性。</p><p>　　移動(dòng)IM應(yīng)用程序：考慮到針對(duì)移動(dòng)

80、IM應(yīng)用程序的研究，在[16]和[17]中模擬用戶在IM應(yīng)用程序上的停留時(shí)間和特有的消息到達(dá)率。 基于這些模型，他們得出了IM應(yīng)用程序的能耗模型。但是，提供的模型僅提供高級(jí)粗粒度行為分析，該分析獨(dú)立于底層IM應(yīng)用程序的操作。 在本文中，我們的重點(diǎn)是在不同IM應(yīng)用程序上進(jìn)行操作。 在其他相關(guān)工作[18]中，作者表明通過(guò)消息捆綁可以減少I(mǎi)M應(yīng)用程序的能耗。為了評(píng)估它們的捆綁算法，作者實(shí)現(xiàn)了一個(gè)定制的IM應(yīng)用程序，并開(kāi)發(fā)了一個(gè)軟件工具Ener

81、gy Box，通過(guò)分析tcpdump跟蹤來(lái)估算發(fā)送/接收即時(shí)消息的能耗。 請(qǐng)注意，提高IM應(yīng)用程序能耗效率的此類技術(shù)有助于我們量化主流IM應(yīng)用程序的能耗。</p><p><b>　　6。結(jié)論</b></p><p>　　通過(guò)將IM應(yīng)用程序的操作分解為6個(gè)狀態(tài)，我們描述了IM應(yīng)用程序的能耗和帶寬效率。 我們還分析了IM應(yīng)用程序的各種操作，例如輸入通知、讀取通知、發(fā)送/

82、接收消息。 我們的分析結(jié)果表明IM應(yīng)用程序中仍然需要進(jìn)行大量改進(jìn)，特別是在“對(duì)話”和“后臺(tái)接收”狀態(tài)下，要提高其能耗效率和帶寬效率。但是，我們觀察到這些應(yīng)用在背景空閑狀態(tài)下已經(jīng)具有相對(duì)較高的能耗效率和帶寬效率。</p><p>　　References</p><p>　　1. AppBrain, http://www.appbrain.com/stats/.</p><

83、;p>　　2. C. Clifford, “Top 10 apps for instant messaging,” Entrepreneur, Dec 11, 2013.</p><p>　　3. BBC, http://www.bbc.com/news/business-22334338.</p><p>　　4. H. Falaki, D. Lymberopoulos, R. M

84、ahajan, S. Kandula, and D. Estrin, “A first look at traffic on smartphones,” ser. IMC ’2010, pp. 281–287.</p><p>　　5. S.-W. Lee, J.-S. Park, H.-S. Lee, and M.-S. Kim, “A study on smart-phone traffic analysis

85、,” ser. APNOMS'2011, pp. 1 – 7.</p><p>　　6. Q. Xu, J. Erman, A. Gerber, Z. Mao, J. Pang, and S. Venkataraman, “Identifying diverse usage behaviors of smartphone apps,” ser. IMC’2011, pp. 329–344.</p&g

86、t;<p>　　7. X. Wei, L. Gomez, I. Neamtiu, and M. Faloutsos, “Profiledroid: multi-layer profiling of android applications,” ser. Mobicom’2012.</p><p>　　8. C. Haas and P. Takayoshi, “Young people's e

87、veryday literacies: The language fea- tures of instant messaging,” Research in the Teaching of English, vol. 45, no. 4, pp. 378–404, May, 2011 .</p><p>　　9. I. A. N. A. (IANA), https://www.iana.org/assignmen

88、ts/.</p><p>　　10. VRMLSite, http://www.vrmlsite.com.</p><p>　　11. Z. Xiao, L. Guo, and J. Tracey, “Understanding instant messaging traffic charac- teristics,” ser. ICDCS’2007, pp. 51–51.</p&g

89、t;<p>　　12. E. Isaacs, A. Walendowski, S. Whittaker, D. J. Schiano, and C. Kamm, “The character, functions, and styles of instant messaging in the workplace,” ser. CSCW ’2002, pp. 11–20.</p><p>　　13.

90、F. Qian, Z. Wang, A. Gerber, Z. Mao, S. Sen, and O. Spatscheck, “Profiling resource usage for mobile applications: a cross-layer approach,” ser. MobiSys '11. ACM, 2011, pp. 321–334.</p><p>　　14. A. Patha

91、k, Y. C. Hu, and M. Zhang, “Where is the energy spent inside my app?: fine grained energy accounting on smartphones with eprof,” ser. EuroSys ’2012, pp. 29–42.</p><p>　　15. L. Zhang, C. Stover, A. Lins, C. B

92、uckley, and P. Mohapatra, “Characterizing mobile open apis in smartphone apps,” in IFIP Networking Conference’ 2014, pp. 1–9.</p><p>　　16. Y. W. Chung, “Investigation of energy consumption of mobile station

93、for instant messaging services,” ser. ISADS’2011, pp. 343–346.</p><p>　　17. ——, “An improved energy saving scheme for instant messaging services,” ser. WiAd’2011, pp. 278–282.</p><p>　　18. E. J.