應急行動通訊系統設計 - 政大學術集成

全文

(1)國立政治大學資訊科學研究所 Department of Computer Science National Chengchi University 博士學位論文 Doctoral Dissertation. 政治大. 立應急行動通訊系統設計. ‧ 國. 學. Design of Contingency Cellular Network. ‧ er. io. sit. y. Nat. al. n. v i n Ch 指導教授：連耀南 e n g c h i U博士博士生：黃智賢撰. 中華民國一百零六年一月 January 2017.

(2) 應急行動通訊系統設計 Design of Contingency Cellular Network. 研究生：黃智賢指導教授：連耀南. Student：Jyh-Shyan Huang Advisor：Yao-Nan Lien. 國立政治大學治. 政. 研究所. 學. ‧ 國. 立資訊科學. 大. 博士論文. io. sit. y. ‧. Nat. A Doctoral Dissertation. er. submitted to Department of Computer Science. al. n. v i n C hChengchi University National engchi U. in partial fulfillment of the Requirements for the degree of Ph.D. in Computer Science. 中華民國一百零六年一月 January 2017.

(3) 誌謝辭在堅持了 9 年半，博士生生涯終於邁入最後關頭，回首這段尋夢踏實的過程，很辛苦，卻也很踏實，在進入職場多年後，還能夠重回校園學習，實在是難得的經驗。這本論文的完成，首先要感謝的是指導老師連耀南教授。老師實事求是且觀察入微的精闢見解與嚴謹的治學態度，讓我不但在學術研究上學習到應有的思考方式，也在相關工作領域上獲益良多。另外感謝逢甲大學黃秋煌教授、台灣大學蔡志宏教授、清. 政治大. 華大學黃能富教授、政治大學蔡子傑教授在口試時的指導與建議，讓這本論文更臻完. 立. 整，在此獻上由衷的感激。此外，還要感謝碩士班的指導教授張宏慶教授，您是我學. ‧ 國. 學. 術研究的啟蒙導師，也是指引我人生方向的明燈，是我的老師，也是最好的朋友。感謝實驗室的學弟妹們，雲鼎、彥嵩、筱慈、冠傑、郁翔、耿誠、采衣、文晟、期. ‧. 淳、惠晴，謝謝你們在研究過程中提出各種看法與建議。. Nat. sit. y. 兼顧工作與求學的過程是艱辛的，要感謝父母親對我的養育與關懷，也要特別感謝. n. al. er. io. 愛妻致蓉，在我博士進修期間對家庭的付出與照顧，以及對 Gloria 和 Deborah 的悉心. i n U. v. 照料與關懷，使我能全心全意的投入工作與學業，謹以此論文獻給她，並致上我最誠摯的謝意與感激。. Ch. engchi. 感謝我的神，祂在過程中一直與我同在，作我最堅強的依靠。最後僅以一句聖經節以賽亞書 30:15「得救在於歸回安息；得力在於平靜信靠」自勉，這一句話在許多的時候給我力量，讓我堅持下去的勇氣。希望在畢業後，能獻上更多的時間和精力給我最愛的主、照顧家人並服事聖徒。黃智賢謹誌於政治大學資訊科學研究所 January 2017. i.

(4) 應急行動通訊系統設計摘要當大型災害來臨，通訊系統對救災效益具有不可或缺的重要性。然而，一般公眾通訊系統，如行動通訊網路等，常因各種不同因素導致系統癱瘓。在災變初期，外援常因交通運輸問題無法即時進駐災區，而需仰頼數以萬計甚至數十萬的災區在地志工投入救. 政治大. 災工作。而現存多個緊急通訊系統的佈建，需仰頼良好的交通運輸，不幸的是，部分道. 立. 路和橋樑常因大型災害而斷裂或變型，導致災區對外交通運輸中斷，無法快速的將緊急. ‧ 國. 學. 通訊系統的網路元件，運送至災區佈建。再者，因造價之故，系統容量僅能提供小部分專業救災團隊使用，而無法提供給數量龐大的救災志工使用。. ‧. Nat. sit. y. 我們提出應急行動通訊系統 (Contingency Cellular Network, CCN) 以提供災區救災工作. n. al. er. io. 的緊急通訊。我們在歷年的災害中發現行動通信系統斷訊時大部分基地台的結構是完整. i n U. v. 的，但因失去與核心網路連線能力或電力供應，而無法提供服務，成為孤立基地台。應. Ch. engchi. 急行動通訊網路(CCN)利用這些孤立基地台搭配無線通訊與衛星通訊技術建置一多重跳接無線網路，以恢復孤立基地台與核心網路連線能力；並配備發電機，提供電力，使孤立基地台可提供有限的服務。救災志工和災民無需使用特殊手持設備或額外的訓練，只需使用原有的手機，即可使用 CCN 的應急通訊服務。CCN 可於第一時間，提供大批救災志工和災民通訊服務，以提高救災效益，因而拯救更多寶貴的生命。. 本論文主要聚焦在應急行動通訊系統設計所衍生出的相關議題，如應急網路需求分析、系統架構設計、網路拓樸規劃、網路頻寬規劃、佈署行程規劃等議題。本論文針對網路拓樸規劃、網路頻寬規劃、佈署行程規劃問題以數學模式進行塑模，並證明這些問題為 ii.

(5) NP-Hard 問題。因網路拓樸規劃、網路頻寬規劃、佈署行程規劃需緊急完成，我們也提出啟發式演算法快速解決這些規劃問題。實驗結果顯示，這些啟發式演算法均具良好的效能。. 關鍵字 — 災難管理、緊急通訊、行動通訊、 Ad Hoc 網路、網路拓樸、佈署行程、頻寬管理. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. iii. i n U. v.

(6) Design of Contingency Cellular Network Abstract Communication system is crucial to the efficiency of disaster response operation in the large-scale disaster. However, communication systems, such as cellular networks, usually crashed due to various causes making coordination among disorganized disaster responders. 政治大. extremely difficult. Unfortunately, rapid deployment of many existing emergency. 立. communication systems relies on a good transportation system, which is usually not available. ‧ 國. 學. in a catastrophic natural disaster.. ‧. We proposed a Contingency Cellular Network (CCN) for emergency communication by. y. Nat. sit. connecting disconnected base stations together with wireless links to construct a wireless. n. al. er. io. multi-hop cellular network. CCN can support existing mobile phone users with reduced. i n U. v. capability. Such a system can support a large number of disaster responders in the early hours. Ch. engchi. of a catastrophic natural disaster, thus save many lives.. Our research addresses various design issues of CCN, such as network topology planning, bandwidth management, deployment scheduling, etc. We take the level of emergency and population of each stricken area as the priority measure as well as the available resources as the constraint to maximize disaster response efficiency. Mathematical models of these design issues are proposed and proved as NP-Hard problems. Since the network topology, bandwidth management, deployment scheduling are needed in urgent, we propose heuristic algorithms to solve these problems quickly. Finally, we evaluated the proposed algorithms by simulation. A iv.

(7) significant improvement in resiliency is reached.. Keywords — Disaster Management, Emergency Communication, Mobile Communication, Ad Hoc Network, Network Topology, Deployment Scheduling, Bandwidth Management. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. v. i n U. v.

(8) 目錄誌謝辭 ......................................................................................................................................... i 摘要 ............................................................................................................................................ ii Abstract ...................................................................................................................................... iv 目錄 ........................................................................................................................................... vi 圖目錄 .....................................................................................................................................viii 表目錄 ....................................................................................................................................... ix Chapter 1. 1.1. 1.2.. Introduction ........................................................................................................ 1. 政治大 Lessons Learn from Haiti and Chi-Chi Earthquake ........................................... 2 立 Disaster Response for Large Scale Natural Disasters ........................................ 1. Robustness Analysis of Cellular Networks in a Disaster ................................... 4. 1.4.. System Requirements for RDN .......................................................................... 7. ‧ 國. Related Works ................................................................................................... 12. ‧. Chapter 2.. 學. 1.3.. Traditional Communication System for Disaster Response ............................. 12. 2.2.. Wireless Network Approaches ......................................................................... 13. y. sit. io. Contingency Cellular Network ......................................................................... 21. n. al. er. Chapter 3.. Nat. 2.1.. i n U. v. 3.1.. Design Philosophy ............................................................................................ 21. 3.2.. System Architecture of CCN ............................................................................ 22. 3.3.. Software Architecture of EC Module ............................................................... 26. 3.4.. Deployment and Operation Procedures ............................................................ 30. 3.5.. Basic CCN Services ......................................................................................... 31. 3.6.. Design Issues of the CCN................................................................................. 32. Chapter 4.. Ch. engchi. Network Topology Design ................................................................................ 37. 4.1.. Considerations of Network Topology Design .................................................. 37. 4.2.. Related Works ................................................................................................... 41. 4.3.. Comprehensive Mathematical Model for Network Topology Design ............. 42 vi.

(9) 4.4.. Complexity Analysis ........................................................................................ 47. 4.5.. Topology Design Heuristic Algorithm ............................................................. 50. 4.6.. Binary Integer Linear Programming Transformation Methodology ................ 54. 4.7.. Performance Evaluation ................................................................................... 64. Chapter 5.. Deployment Scheduling ................................................................................... 76. 5.1.. Considerations of Deployment Scheduling ...................................................... 76. 5.2.. Related Works ................................................................................................... 77. 5.3.. Resource Delivery Path Dependent Deployment Scheduling .......................... 78. 5.4.. 政治大 Complexity Analysis ........................................................................................ 81 立 Heuristic DS-ACG Algorithm .......................................................................... 84. 5.6.. Heuristic DS-UCB Algorithm .......................................................................... 84. 5.7.. Performance Evaluation ................................................................................... 85. ‧ 國. ‧. Bandwidth Allocation ....................................................................................... 89. y. Nat. Chapter 6.. 學. 5.5.. 6.2.. Bandwidth Allocation Mathematical Model..................................................... 89. 6.3.. Complexity Analysis ........................................................................................ 91. 6.4.. CCN Bandwidth Allocation Heuristic Algorithm ............................................. 93. 6.5.. Performance Evaluation ................................................................................... 96. n. al. er. sit. Considerations of Bandwidth Allocation..........................................................89. io. 6.1.. Ch. engchi. i n U. v. Chapter 7. Conclusion ...................................................................................................... 100 References .............................................................................................................................. 103 Appendix I : Pseudo Code of Total profit maximization Process .......................................... 109 Appendix II : Pseudo Code of Binary Integer Linear Programming Algorithm .................... 112 Appendix III : Pseudo Code of Heuristic DS-ACG Algorithm .............................................. 122 Appendix IV : Pseudo Code of Heuristic DS-UCB Algorithm .............................................. 124. vii.

(10) 圖目錄 Figure 1.1. Survival rate statistics .............................................................................................. 1 Figure 1.2. A fell building divided a rescue squad into two parts .............................................. 3 Figure 1.3. Broken bridge cut off communications cables ......................................................... 4 Figure 1.4. Vulnerable points of cellular network ...................................................................... 7 Figure 1.5. 7-Ability of rapidly deployable network ................................................................ 10 Figure 2.1. System architecture scheme of the metropolitan area approach ............................ 16 Figure 3.1. System architecture of CCN................................................................................... 24 Figure 3.2. Use scenarios of CCN ............................................................................................ 25 Figure 3.3. Software architecture of EC Module...................................................................... 26 Figure 4.1. Considerations of topoloy design ........................................................................... 38. 立. 政治大. ‧ 國. 學. Figure 4.2. Four types of CCN network topology design problems......................................... 40 Figure 4.3. Main flow of Topology Design Heuristic Algorithm ............................................. 50 Figure 4.4. Flow chart of pivot path construction .................................................................... 52. ‧. Figure 4.5. Flow chart of total profit maximization ................................................................. 53 Figure 4.6. Example of relationship between nodes and edges ................................................ 56 Figure 4.7. Example of relationship between paths and edges ................................................. 57 Figure 4.8. Example of multiple disjoint paths ........................................................................59 Figure 4.9. Characteristics of Simple and Cross FT (optimal solution) ................................... 68 Figure 4.10. Characteristics of MPFN and Cross MPFN (number of pivot nodes/paths) ........ 69 Figure 4.11. Performance of TDHA (deviation of optimal profits)..........................................70 Figure 4.12. Performance analysis of TDHA (Simple and Cross FT) ...................................... 71 Figure 4.13. Performance analysis of TDHA (MPFN and Cross MPFN) ................................ 72. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 4.14. Reliability of multi-path network topology (box plot) ......................................... 73 Figure 4.15. Network reliability and unit profit gain ............................................................... 74 Figure 4.16. Depth weight based vs. depth bound based ......................................................... 75 Figure 5.1. Example of deployment schedule .......................................................................... 79 Figure 5.2. A path ignoring antecessor precedence constraint. ................................................ 81 Figure 5.3. Total profit and total deployment time in experiment I ......................................... 88 Figure 5.4. Total profit and total deployment time in experiment II ........................................ 88 Figure 6.1. Flow chart of BAG algorithm ................................................................................ 95 Figure 6.2. Normalize Deviations of experiment I, II. ............................................................. 98 Figure 6.3. Pseudo deviations of experiment III. ..................................................................... 98 viii.

(11) 表目錄 TABLE 2.1. Comparisons of emergency communication solutions for disaster response ...... 13 TABLE 4.1. Mapping table of two key factors and network topologies .................................. 51 TABLE 4.2. Parameters for experiment setup .......................................................................... 65 TABLE 4.3. Evaluation metrics................................................................................................ 66 TABLE 4.4. Environments of experiments .............................................................................. 66 TABLE 5.1 Parameters of test instances in experiment I ......................................................... 86 TABLE 6.1 Three sets of experiments ..................................................................................... 96 TABLE 6.2 Parameters for test instances ................................................................................. 97. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. ix. i n U. v.

(12) Chapter 1. Introduction 1.1. Disaster Response for Large Scale Natural Disasters Frequently occurring large-scale natural disasters have been reported to cause great damage in recent years, claiming many people’s lives, rendering millions of people homeless,. 治政 and causing a huge financial loss. The earthquake that occurred 大 in Haiti in 2010 alone claimed 立 230,000 people’s lives. When a large-scale disaster strikes, the destroyed areas are often sunk ‧ 國. 學. into desperation. Take the earthquake that happened on March 11, 2011 in North-Eastern. ‧. Japan for example, after stricken by a 9.0 magnitude earthquake, followed by 23-meter high tsunami and the combined major disasters (i.e. nuclear crisis, earthquakes and tsunami), the. y. Nat. n. al. er. io. disaster.. sit. world and the already experienced disaster responders were all stunned by the huge scale of. Ch. engchi. i n U. v. Figure 1.1. Survival rate statistics 1.

(13) One major task for an earthquake response is to rescue victims who are trapped in the rubble of collapsed buildings. The survival rates are highly depended on the time that the victims are rescued. Survival rate statistics is showed in Fig. 1.1. The survival rate may be as high as 90% in the first 24 hours, but may drop to 55% and 25% after 48 and 72 hours, receptivity. To rescue trapped and wounded victims within Golden 72 Hours is a critical task of Disaster Response Operation (DRO).. 治 and Chi-Chi Earthquake 1.2. Lessons Learn from 政 Haiti 大. 立. Haiti Earthquake occurred in January 2010, claimed more than 230,000 lives. Entire. ‧ 國. 學. government and social infrastructure were completely paralyzed. All utilities such as running water, electricity, communications, food, and medical supplies were all knocked out.. ‧. Port-au-Prince Airport was forced to shut down because it could not supply fuel for airplanes. y. Nat. sit. to take off. Most roads were also destroyed. Therefore, DRO resources and equipment were. n. al. er. io. extremely difficult to deliver to the stricken zones. In the first few days, Haitian people had to. i n U. v. help themselves with little external aid. Internet was lucky to survive the quake because most. Ch. engchi. ISPs relied on satellite for their external connections.. Chi-Chi/Taiwan. Earthquake. occurred. in. September. 1999,. claimed. 2,415. lives.. Communication systems were mostly down or jammed resulting in big impact to DRO [13]. Chunghwa Telecom took 15 days of 7/24 operation to restore its systems. A large number of local volunteers were mobilized to execute DRO and it’s difficult to coordinate large number of voluntary responders without a good communication system.. 2.

(14) 立. 政治大. ‧. ‧ 國. 學. Figure 1.2. A fell building divided a rescue squad into two parts. y. Nat. sit. Without a good communication system may cause rescue efforts interfere with each other. For. n. al. er. io. instance, in Chi-Chi Earthquake, sound detection operations were interfered by sirens. The. i n U. v. collapsed buildings as showed in Fig. 1.2 fell down and blocked the street dividing first. Ch. engchi. responders into two isolated groups. While one group was doing sound-sensitive operation (e.g. using a high sensitive sound detector to detect any human sound under debris), the group on the other side was using heavy machinery to dig the rubbles. It also happened that rescued victims died on the way being transferred from hospital to hospital. It is because Victim Arrangement System (ambulance) couldn't follow up changing status of hospital capacities. Some rescue and relief resources were misplaced because assessment of disaster distribution was blind and inaccurate. Reallocation of misplaced resources may not be possible due to paralyzed transportation systems. Misplacement of rescue and relief resources may lead to catastrophic consequence. 3.

(15) Communication systems are critical in disaster response operations. However, they are not dependable, including cellular networks. Any emergency communication system proposal must consider (1) large number of disorganized voluntary responders and victims, (2) transportation system may be paralyzed, (3) external aids may not be available in Golden 72 Hours.. 治 Networks in a Disaster 1.3. Robustness Analysis 政of Cellular 大. 立. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 1.3. Broken bridge cut off communications cables Public communication and transportation systems are often completely or partially destroyed after large-scale natural disasters [13,23,45], and thus transporting resources into disaster zones to restore communication systems is difficult. There may be more than ten thousand or one hundred thousand of local volunteers are mobilized to execute DRO. And, it is hard to coordinate these disorganized voluntary responders. Hence, following a disaster, the 4.

(16) first responders demand a rapidly deployable network (RDN) to provide connectivity. Communication systems are known to be crucial to disaster response; however, apparently stable public communication networks frequently do not survive natural disasters. Notably, cell phone networks were vulnerable during the 88 Flood and 921 Chi-Chi Earthquake [13] in Taiwan. When Hurricane Sandy hit the East Coast of the United States, one quarter of base stations in the East Coast were disrupted. The following reasons for the vulnerability were later identified: [23]. 政治大 Common causes for service disruption of base stations are: (1) power outage (the backup 立 batteries usually can only last several hours), (2) broken backhaul, and (3) physical. ‧ 國. 學. destruction by disasters. Critical hardware was knocked out resulting from (1) external power. sit. y. Nat. switches caused by the loss of cooling systems.. ‧. outage, (2) exhausted power generator fuel, (3) broken cooling systems, or (4) overheated. io. er. Because base stations must be connected to controllers or switches through backhaul cables,. al. v i n C hpower lines and backhaul structure remains intact. Typically, e n g c h i U links are laid along roads and n. they cannot remain in operation when their backhaul is disconnected, even if their physical. bridges for the convenience of deployment and maintenance. The destruction of roads and bridges is a common phenomenon in disasters and leads to power outages and network disconnection (see Fig. 1.3, 1.4). Although power lines and communication backhauls usually have inbuilt redundancy to increase availability, they do not necessarily improve survivability significantly in large-scale disasters. For instance, a substantial flood may destroy many bridges over a river, and thus completely break all of the redundant cables. In the 88 Flood, the structure of many base stations remained intact because they were located in higher places. However, when the power lines and backhauls that were laid along the roads and bridges were 5.

(17) destroyed by the flood, mobile communications systems were paralyzed. In short, power lines and backhauls are an Achilles’ heel for many existing mobile communication networks. Because of the aforementioned problems, building stronger mobile communication networks has been proposed. However, such robust networks are prohibitively expensive for large-scale deployment. For instance, although the National Communications Commission of Taiwan selected a small number of locations in which to build strengthened base stations with satellite communication for backhauls, the number of such base stations is limited due to funding constraints.. 立. 政治大. Long Term Evolution (LTE) is a network technology that will be used for emergency. ‧ 國. 學. communications in the future. Currently, Canadian public safety agencies are attempting to. ‧. clear up new spectrum bands, leverage commercial services, and reuse existing spectrum. sit. y. Nat. assets. On June 11, 2012, Trans-European Trunked Radio (TETRA) and Critical. io. er. Communications Association (TCCA) and National Public Safety Telecommunications Council (NPSTC) announced that they had signed a Memorandum of Agreement to. al. n. v i n commitment mission-critical C h to developing engchi U. underscore. their. joint. public. safety. communication standards for LTE-based technology [14,18,19,54]. However, key features must still be implemented in LTE before it can satisfy public safety requirements. Before the LTE is applied in disaster response, a rapidly deployable network (RDN) is necessary to assist first responders. A RDN is an adaptive, mobile communications network that can be easily accessed. Its purpose is to temporarily replace damaged public networks after a disaster, thereby providing emergency communications among the first responders and, if possible, the victims. As the permanent communications for the affected population are gradually restored, the RDN is 6.

(18) withdrawn.. 立. 政治大. ‧. ‧ 國. 學 er. io. sit. y. Nat. Figure 1.4. Vulnerable points of cellular network. 1.4. aSystem Requirements for RDN. n. iv l C n e n g cobtained On the basis of our firsthand h experience h i U in the 921 Chi-Chi Earthquake and extensive research over the past decade [14,13,17,23,42]. We summarized the environmental constraints in the disaster areas are described as follows, (1) large number of disorganized non-professional voluntary responders and victims, (2) transportation system paralyzed, (3) time is running out, (4) hectic/chaotic usage environment, and (5) very limited deployment funding due to little commercial incentive.. We also summarized a set of communication requirements that must be addressed when. 7.

(19) constructing and operating a communication network for first responders. These requirements are categorized into two sets: user end and operator end.. User End Requirements Popularity: In large-scale disasters, numerous volunteers must be mobilized to work on rescue and relief operations. In addition, people in the disaster area—including victims—may. 政治大 for a RDN. Because of the rareness of terminals, most common emergency communication 立. have extensive communication requirements. Therefore, several user terminals are necessary. ‧ 國. 學. networks, such as satellite systems, trunking radios, and amateur radios, can only be used by specific groups; most victims and volunteer disaster responders cannot access these. ‧. communication networks. Moreover, users must be trained before using the terminals for. sit. y. Nat. trunking radio and amateur radio; hence, these systems can only be employed by professional. io. n. al. er. disaster response squads.. Ch. engchi. i n U. v. Usability: A RDN should provide task-oriented communication services and support mobility, as well as have adequate service quality. Furthermore, RDN handsets should be user-friendly, durable, and not require a long training period, and task-oriented communication services should include both ordinary and group communication services. Finally, because disaster responders may have to move frequently, the mobility of user terminals is crucial.. Operator End Requirements Practicality: Practicality is the essential operator end requirement, and includes low 8.

(20) deployment costs, easy acquisition of equipment, and rapid deployment. Although RDNs are essential to disaster response operation, they do not generate profit and the occasions for their use are limited. A commercial cell phone operator may not be justified in making a large investment in the design and development of a large-scale RDN; therefore, a RDN must have low development costs to be practical. Because of obstacles created by terrain and paralyzed transportation systems, transporting external aid to disaster areas is usually difficult. In many cases, helicopters may be the only. 政治大. vehicle that can access disaster areas in the early hours or days of a large-scale disaster. Hence,. 立. the size and weight of RDN equipment should be suitable for air transportation. For instance,. ‧ 國. 學. mobile base stations (called “cells-on-wheels”), which have a base station with satellite. ‧. backhaul and are carried by a truck, may be too heavy to be carried by a helicopter; hence, this equipment may be useless.. sit. y. Nat. io. er. Finally, survival rates are highly dependent on rescue speed. If trapped victims are rescued quickly, their chances of survival are considerably higher; thus, RDNs should be deployed as. al. n. v i n swiftly as possible. Furthermore,Ccell phone operators must work at full capacity to restore hengchi U. their systems. The value of a “band-aid” style RDN is substantially lower once any cell phone is recovered. Therefore, a RDN must be rapidly deployable.. Capacity: A RDN must have sufficient capacity to satisfy the communication demands of large numbers of victims and disaster responders—both professional and voluntary—within stricken areas, as well as limited incoming and outgoing calls to external institutions. Furthermore, a RDN should have the ability to resist the burst of call requests, to prevent it from crashing. 9.

(21) Sustainability: A RDN should not only be deployed as quickly as possible but should also continuously operate until the public communication network is recovered, which may take several days or even weeks. If nonstop operation is impossible, it should be rapidly recoverable once it has crashed.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 1.5. 7-Ability of rapidly deployable network Adaptability: Similar to a battle field, the situation in a disaster area may constantly change due to factors such as aftershocks, fires, and the progress of disaster response. Therefore, a RDN must be able to adapt to the changing environment either manually or automatically. 10.

(22) Operability: Similar to any production system, a RDN must have an operation, administration, and maintenance function to remain in operation.. The system requirements are summarized as 7-ability of rapidly deployable network and showed in Fig. 1.5.. The layout of this. 政治大 dissertation is as follows. Chapter 立. 2 introduces the traditional. ‧ 國. 學. communication systems for disaster response and current available solutions with a comparative evaluation against the 7-ability requirement. We design a large-scale, low cost. ‧. and rapidly deployable communication system called Contingency Cellular Network (CCN). sit. y. Nat. to support large number of disorganized users. Chapter 3 discusses the design philosophy,. io. er. system architecture and research issues of CCN. Chapter 4 formulates the network design. al. v i n C h are proposed to Usolve the problem. Effectiveness and topology design heuristic algorithm engchi n. problems of CCN with a comprehensive mathematic model. A binary linear algorithm and a. efficiency of the algorithms are verified by simulation. Chapter 5 formulates the deployment scheduling problem. Two heuristic algorithms are proposed and discussed. Chapter 6 formulates the bandwidth allocation problem and proposes a solution. Finally, Chapter 7 concludes this dissertation with a discussion of the contribution of our research works.. 11.

(23) Chapter 2. Related Works 2.1. Traditional Communication System for Disaster Response Conventional communication systems that have been widely used in disaster responses are Walkie-Talkie, amateur radio, trunking radio, mobile satellite phone, and Cell-on-Wheel. 政治大 by many researchers. A MANET based system uses mobile computing devices such as laptop 立. (mobile base station), etc. Recently, WiFi based Ad Hoc network (MANET) has been studied. ‧ 國. 學. PC, tablet PCs and smartphones to construct an emergency communication network. Some may have satellite connection to the external Internet. Users can use a mobile device running. ‧. a VoIP application to access the communication service [34]. A brief comparison of these. sit. y. Nat. technologies using the 7-ability requirement presented in Chapter 1 is shown in Table 2.1.. io. n. al. er. Each of them has its own advantages and limitations.. v. Walkie-Talkie, amateur radio, satellite communication and trucking radio are often used in. Ch. engchi. i n U. disaster response. All of them need special user terminals. They have high usability and practicality, but low popularity. Except a few countries such as the United States, Walkie-Talkie is not very popular among ordinary people. According to our experience in 88 Flood, it took Taiwanese government more than two weeks to borrow approximately 1000 Walkie-Talkie sets from vendors to support disaster response. Nowadays, the popularity of Walkie-Talkie even in the United States is much lower than that of cell phone. Cell-on-Wheel can be deployed rapidly to the disaster area to support cell phone users. However, due to its high cost, local cell phone operators may not have a sufficient number of such equipment ready for a large-scale natural disaster. Furthermore, a Cell-on-Wheel system 12.

(24) is usually built on a truck such that it may have difficulty to be transported to the afflicted areas from either local or foreign areas. TABLE 2.1. Comparisons of emergency communication solutions for disaster response Practicality. Walkie-. Usability. Terminal. Terminal. Terminal. Popularity. Usability. Mobility. Low to. Low. High. Moderate. learning. Talkie. 立. cost. Radio. Profession. Moderate. Moderate. Capacity. Per User. Deployment. Transportation. Concurrent. Cost. Difficulty. Demand. User Limit. Low. None. 政治大. Low. ‧ 國. Low. Quality. Moderate. Professional. al skill. Low. No. of handsets. 學. Amateur. Popularity. Low. skill required. No. of handsets. required High. Moderate. Low. learning cost. Cell-On-. High. Easy. Low. No. of handsets. y High. al. High. High. n. Radio. Low. io. Trunking. None. sit. Phone. Very High. Nat. Mobile. Easy. Ch. High. engchi High. i n U. High. Easy. er. Low. ‧. Satellite. High. v. No. of handsets. Easy. Wheel. High. No. of Cell-on-Whe els. 2.2. Wireless Network Approaches Numerous researchers have focused on the approach to RDNs [17,42]. Several approaches to RDNs, including examples, are introduced in this section. An essential set of challenges must be conquered when deploying a network under critical conditions [23,42]. 13.

(25) First, the RDN must be deployed with limited prior knowledge of the environment and as efficiently as possible to replace the damaged portion of the public communication network. However, because traffic movement is paralyzed in most scenarios, transporting professionals and network components into the disaster zone is difficult. The RDN must reuse the network components in the disaster zone and be deployed with minimum human intervention. Second, the network must be adaptive, self-reconfigurable, flexible, scalable, and energy efficient to cope with unknown dynamic environments and battery-powered wireless devices. In other. 政治大. words, the network must be set up on demand in accordance with the location and current. 立. needs.. ‧ 國. 學 ‧. The mobile ad hoc network approach uses on-hand mobile devices to construct a mobile ad. sit. y. Nat. hoc network (MANET) and provide connectivity among the first responders in the absence of. io. er. external network devices [34]. This approach uses the mobile devices of participants, such as smartphones and notebooks, to construct a MANET. These participant nodes build. al. n. v i n interconnections by using wireless C htechnologies, suchUas IEEE 802.11x and Bluetooth, and engchi cooperatively routes packages. Some participant nodes may have satellite connections, enabling a MANET to connect to the Internet. In the case of failure, the packages can be sent by selecting an alternating forwarding path. MANET is flexible and rapidly deployable, its distribution requirements are simple, and it can rely entirely on first responders’ mobile devices without external network devices; hence, it can be applied in most disaster zones. However, its coverage area and support for real time communications are limited. MANET is most suitable for small scale areas, and helping the. first responders share text messages, pictures, and videos for rescue purposes. Examples of 14.

(26) MANET are described in the following section. Lien [36] used notebooks to construct a MANET and developed a Voice over Internet Protocol (VoIP) application to provide a walkie-talkie-like communication service. Notably, the audio quality decreases as the number of hops increases, and the experimental results indicated that this is because the delay time and distortion become longer and more unstable when the number of hops increases. When the number of hops exceeds four, the audio is entirely blurred. In other words, the coverage range is limited when using a MANET to. 政治大. provide a voice communication service.. 立. Bruno [8] proposed the use of wireless mesh networks, in which a set of nodes is dedicated to. ‧ 國. 學. forwarding the traffic of the other nodes. These dedicated nodes form a multi-hop wireless. ‧. backhaul to extend coverage and enhance network performance.. sit. y. Nat. MANET based systems can support laptop and smartphone users. They are easy to construct. io. er. in terms of physical connections and can use user’s own equipment such that it doesn’t count. al. v i n C hlink and VoIP overUMANET may cause a long delay time one critical reason: a GEO satellite engchi n. on pre-allocated funding to acquire user end devices. However, their quality is skeptical for. which may severely hurt the quality of a phone conversation. Finally, MANET is not a commercially mature product and may not be able to obtain sufficient financial support for further research and development due to the lack of commercial incentive.. 15.

(27) 立. 政治大. ‧. ‧ 國. 學. Figure 2.1. System architecture scheme of the metropolitan area approach. The metropolitan area approach aims to deploy network components to cover large incident. y. Nat. io. sit. areas. System architecture scheme of the metropolitan area approach are showed in Fig. 2.1.. n. al. er. This approach applies a layer infrastructure to form a wireless mesh backbone. The first. Ch. i n U. v. responders connect to base stations or access points that are mostly carried by vehicles, which. engchi. connect to the Internet through an aerial or aerospace node. The systems of the metropolitan area approach cover a larger area and provide superior network performance to those of MANET approach; their ability to support real time services is also better. However, the network components of these systems require adequate transportation to be transferred into an incident area and transportation systems are usually paralyzed after large-scale nature disasters. In addition, the system architecture is more complicated and requires professionals to be deployed. Examples of metropolitan area systems are described in the remainder of this section.. 16.

(28) Aerospace Communications for Emergency Applications [7] adopts a layered composition. The first responders connect to base stations or access points that are mostly carried by vehicles, such as ambulances, police cars, or fire trucks. Base stations are connected to an aerial or aerospace node, and the access points use virtual cell layout (VCL) resources to create an overlaid real cell that moves over the virtual cells. Man-pack radios (MPRs) form small cells, each of which has a node that plays the role of an MPR cluster head, and are grouped into a larger cell through radio access points. Finally, all cells are connected through. 政治大. a satellite or an unmanned aerial vehicle. The VCL approach focuses on adapting 3G. 立. technologies to tactical systems.. ‧ 國. 學. The CHORIST Broadband and Wideband Rapidly Deployable Systems [1] involve forming a. ‧. mesh network using wireless routers carried by emergency vehicles. The vehicles are automatically connected in a peer-to-peer fashion and form a self-configuring intervehicular. y. Nat. io. sit. core. At the edges, the mobile radios carried by the first responders are connected to the. n. al. er. closest router via WiFi and create local cells; the remote connection to the command center is. Ch. established through an IP backbone.. engchi. i n U. v. Advanced Wireless Ad-Hoc Networks for Public Safety [2] is a cross-layer approach coupled with a cluster-based architecture that provides a high bitrate service using ad hoc hot spots, and includes PHY, MAC, and network levels. The physical layer relies on reconfigurable Orthogonal Frequency-Division Multiple Access with multi-antenna capability (multiple input and multiple output) to provide data transport services. In addition, this approach includes a MAC layer organized in clusters wherein, in a given set of nodes, the cluster head is the optimally located node. Relay nodes are used to ensure interconnectivity between clusters, and gateway nodes are used to connect the created network to other networks. The 17.

(29) cornerstone of this proposal is the terminode concept, for which each node in the network is able to perform the functions of a cluster head, relay, router, or gateway, depending on its location and service requirements. The application of 3G PCS Technologies to Rapidly Deployable Mobile Networks [9] was proposed for military communications. According to the cluster-based organization in a layered architecture, this approach provides a cluster head for each group of first responders. The cluster head acts as a bridge between the first responders and the access points carried by. 政治大. the vehicles, and there is at least one cluster head connecting the remainder of the nodes to an. 立. upper layer, such as a satellite acting as a gateway.. ‧ 國. 學. The Satellite-Assisted Localization and Communication Systems for Emergency Services [46]. ‧. involves an integrated navigation/communication (NAV/COM) reconfigurable system. sit. y. Nat. architecture for emergency scenarios. In this architecture, rescuers within the incident area. io. er. network are organized into teams or clusters. The main topics of the project are localization techniques, software-defined radio, cognitive radio NAV/COM devices, the integration of. al. n. v i n satellites and high-altitude platforms and the adoption of heterogeneous C hinto rescue services, engchi U solutions in the intervention area.. The local area approach aims to deploy a network to a target area using small, inexpensive, and light-weight wireless relay nodes. An on-demand multi-hop wireless forwarding network is created that allows the first responders to communicate as soon as the relay nodes are deployed. The primary problem with this approach is identifying an appropriate deployment decision algorithm to maximize network performance. On the basis of the deployment strategies of the relay nodes, this approach is divided into two categories, namely the static 18.

(30) relay nodes approach and the automobile relay nodes approach.. In the static relay nodes approach, the relay nodes are mainly deployed by the first responders. The first responders are notified to drop the relay nodes at a certain distance to maintain the connectivity of the relay nodes. The connectivity quality can be measured using a received signal strength indicator (RSSI) [5], the signal-to-noise ratio (SNR) [49] or. 政治大 proposed a deployment decision strategy based on the connectivity quality measured by the 立. bandwidth [53]. Moreover, there are two types of deployment decision strategies. Souryal [49]. relay node. A relay node constantly sends signals to measure the SNRs of neighboring nodes,. ‧ 國. 學. and once the SNR exceeds a predefined threshold, the relay node sends a deployment request. ‧. to the closest first responder. Bao [5] proposed a similar deployment decision strategy based. sit. y. Nat. on the connectivity quality measured by the devices of the first responders. In this approach,. io. er. the first responder or the closest neighbor receives a deployment request if the RSSI between his device and the relay node crosses a predefined threshold.. al. n. v i n Cdeployed Relay nodes are inexpensive and U the coverage area can be simply h e n geasily. c h iHence, extended by dropping more relay nodes into an incident area. However, these relay nodes remain static after deployment and cannot be adjusted according to environmental changes. Moreover, the deployment of relay nodes is reliant on first responders. Once the first responders have entered an incident area, deploying relay nodes is not their primary task, and they may be unable to drop a relay node after receiving a deployment request. This may cause the degradation of network performance or partition of the coverage area.. The automobile relay nodes approach overcomes the weakness of the static relay nodes 19.

(31) approach by reducing the degree of human intervention. Here, the deployment of relay nodes is not reliant on the first responders, because the relay nodes are automobiles that can self-spread across the target area following deployment. There are two types of deployment strategies. The first one involves guiding the automobile relay nodes to form a chain formation [31,43,44], thereby restoring the connectivity between the APs of the mesh network or creating connectivity between the command center and the rescue team. The second [47,50] attempts to cover the target area thoroughly, to connect as many rescue workers and victims as possible.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 20. i n U. v.

(32) Chapter 3.. Contingency Cellular Network 3.1. Design Philosophy. After a decade of study, we have discovered that most base stations crash during disasters because of the damage to their power source or backhaul links. Therefore, we designed a new RDN called the Contingency Cellular Network (CCN) [23,24]. In this. 政治大 intact are connected using wireless links to form a multi-hop cellular network. Traffic can be 立 approach, isolated base stations that have had their service disrupted but remain physically. ‧ 國. 學. forwarded hop by hop from any isolated base station to a survival base station. A base station which has external link to core network is called survival network. The main design concept. ‧. of the CCN was to reuse existing disconnected base stations to minimize cost and deployment. sit. y. Nat. time, and to support a large number of existing users. The rationale for this system was that (a). io. er. mobile communication networks provide wide coverage; (b) cell phone use is widespread; (c). al. only a low-cost add-on module is required to repair a disconnected base station; and (d) there. n. v i n whichCishnecessary in current e n g c h i U disaster response communication.. is a low barrier of use,. Furthermore, cell phones are likely to be carried by affected people in a disaster area (a crucial, although nontechnical justification for this approach). Therefore, reconnecting disconnected base stations in a disaster area to provide a low-cost large-scale emergency communication service is a favorable option. The Contingency Recovery Package (CRP) consists of a power module, several Intercell Communication (ICC) modules, and an add-on processing module, which is referred to as an Emulated Controller (EC) module. The CRP can be stored in national disaster response centers or with cellular network operators and delivered to base stations using airdrops or 21.

(33) helicopters. In the first step, an EC module is connected to a base station. ICC modules are then used to connect the base station to its neighbors using long range wireless links; notably, at least one pair of ICC modules is necessary for each base station. A multi-hop wireless network overlapped on top of the selected base stations is subsequently formed, which provides the connectivity between the base stations and the core network. Anyone who has a cell phone can access service through these base stations. If there is no way to connect to the core network, some CRPs may be equipped with a satellite modem to establish a connection.. 立. 政治大. ‧ 國. 學. 3.2. System Architecture of CCN. The system architecture is showed in Fig. 3.1. The details of power, ICC, and EC. ‧. modules are as follows.. y. Nat. sit. Power module: This module consists of a portable power generator and required fuel that is. n. al. er. io. sufficient to provide electricity to a base station for at least a few days. Note that although. i n U. v. most base stations have backup power, it usually can last only a few hours.. Ch. engchi. Inter-Cell Communications Module (ICC Module): Its main function is to establish connections between base stations. There is usually no wired connection between base stations. Major components are a wireless transceiver and an antenna.. Emulated Controller Module (EC Module): EC Module is the core controlling component of CCN. Its main functionalities are establishing connections between base stations and transferring telecommunication signaling as well as acting as a PBX to provide intra-CCN 22.

(34) communication services. Because the external bandwidth of CCN will be shared by all base stations, there must be many radio channels (from base stations to cell phones) remaining idle. EC Module uses these idle channels to provide intra-CCN communication services. There are many low-cost solutions to implement EC Module. A powerful laptop equipped with an interface to the ICC module and the target base station (most likely an Ethernet interface) running Linux operating system will be sufficient. Note that the majority of traffic, which is intra base station traffic, will be handled by base station itself, but not EC Module.. 政治大. Inter base station traffic will be low because it is regulated by the bottlenecks, the external and. 立. inter base station links.. ‧. ‧ 國. 學. Satellite Modem: This optional module provides the connection between CCN and the core. sit. y. Nat. network once the connections to the core network are all broken. Satellite communications,. io. er. which is not confined by the geographical boundaries, can connect to the core network. al. v i n C h the multi-hop connectivity. connected to the core network through engchi U n. directly. However, only a few base stations can be installed due to its high cost. Others will be. CRP can be stored in national disaster response centers or cellular operators and delivered to the selected base stations via any transportation means even airdrops or helicopters. The EC-Module is connected to a base station in the first step. Then, ICC Modules are used to connect the base station to its neighbors in the second step via long range wireless links. At lease a pair of ICC Modules is needed for each base station. A multi-hop wireless network overlapped on top of the selected base stations is finally formed. The overlapped network. 23.

(35) provides the connectivity between base stations and the core network. Anyone who has a cell phone can access the contingency communication service from these base stations. If the constructed CCN is completely isolated from the core network, some CRP may include a Satellite Communication Module (SC Module) to establish a connection to the core network. In order to maintain the connectivity of base stations, the forwarding tree is re-planned. 政治大. immediately, if any of the links of the forwarding tree is missing.. 立. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 3.1. System architecture of CCN The use scenarios of CCN are illustrated in Fig. 3.2. In Scenario 1, some base stations are out of service, a result of losing connections with the core network; however, their structures are intact. These base stations are called isolated base stations. The CCN recovers the. 24.

(36) functionality of these isolated base stations by reconnecting them to the core network. ICC modules are used to construct wireless links among the base stations and reuse the undamaged backhaul links to form a multi-hop wireless network that recovers connectivity between the isolated base stations and the cellular network. In Scenario 2, the command center is located in an area that requires more bandwidth and higher reliability. The CCN satisfies the requirements by constructing multiple disjointed paths between the command center and the core network.. 政治大 In Scenario 3, when an isolated base station cannot connect to the core network through other 立. base stations, the CCN recovers the backhaul link of this isolated base station by using. ‧ 國. 學. satellite and satellite modem to reconnect to the core network.. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 3.2. Use scenarios of CCN. 25.

(37) 3.3. Software Architecture of EC Module Software architecture of EC Module is showed in Fig. 3.3. Components of EC Module are illustrated as follows:. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. i n U. v. Figure 3.3. Software architecture of EC Module. engchi. Service Process Controller: responsible for service procedure and cooperates with the function components to fulfill the service to the users.. User Profile: records user information such as phone number, IMSI, agency group, service level agreement (SLA), etc.. 26.

(38) Service Definition: the definitions of communication services include service procedure and corresponding functions. New communication is added/modified by adding/modifying its’ service definition.. BS Controller Emulator: deals with the communications protocol with base stations and transfers signal and data into VoIP package. The existing mobile network base stations come. 政治大 update of the base stations, EC Module may not work properly if it is not updated. EC 立 from different companies, and the software and hardware are updated constantly. After the. Module is a kind of emergent equipment and has no ample fund and resources to update. ‧ 國. 學. constantly. To ensure the utility of CCN, we use BS controller emulator to deal with the. ‧. problem of base station connection. When the base station is updated, only the BS controller. n. al. er. io. sit. y. Nat. emulator is required for renewal, with other functions unchanged.. i n U. v. Service Center: supplies the necessary function to provide communication service. Since. Ch. engchi. CCN supports three types of communication modes, service center discriminates the communication mode of an incoming call and uses corresponding function to fulfill user’s request by referring to the service definition.. User InfoCenter: responsible for user identity. There are two kinds of users: one is agency group member and the other is anonymous who has no user information in CCN. When an anonymous user connects to CCN, User InfoCenter records its’ user information with default agency group and SLV.. 27.

(39) QoS manager: is responsible for deciding the QoS of incoming calls. Incoming calls have different level of emergency. Based on the emergency level of a phone call, appropriate QoS will be provided. The more urgent one will be allocated with higher bandwidth to provide better service quality, and vice versa. The emergency level of a phone call is determined by the agency groups registered by both caller and callee. If neither the caller or the callee is registered, the priority will be given to those who have registered. The main purpose of. 政治大. bandwidth differentiation according to the urgency of the phone calls is to answer as many. 立. phone calls as possible without affecting the disaster response efficiency.. ‧. ‧ 國. 學. Mobility Manager: is responsible for the interrogation of callee locations.. sit. y. Nat. n. al. er. io. Network Process Controller: is responsible for network service procedure and cooperates. i n U. with the function components to fulfill network service.. Ch. engchi. v. Disaster Situation Data: records the situation data of disaster areas, such as population distribution, numbers of victims and responses workers. In order to efficiently support responders, CCN collects the disaster situation data and allocates network resources accordingly.. Network Data: records the network related information such as network topology, bandwidth allocation, bandwidth utilization, etc. 28.

(40) Admission policy: records the admission policy of network resources such as inter-BS bandwidth and satellite bandwidth.. Admission Controller: prioritizes urgent phone calls and admits phone calls. The capacity of CCN is much smaller than usual public network, making it difficult to handle massive phone. 政治大 victim rescue. Since CCN capacity cannot provide service to all phone calls, every possible 立. calls. Also, the level of urgency varies, ranging from regular phone calls to urgent calls for. ‧. ‧ 國. in CCN.. 學. means should be made to prioritize urgent phone calls for them to access the telephone service. y. Nat. sit. Routing Manager: is responsible for planning network topology and routing path from. n. al. er. io. source to destination. Network topology is planned in planning phase firstly. And then,. i n U. v. network topology is dynamically re-planned according to the change of communication. Ch. engchi. requirement to maximize the efficiency of disaster response.. Bandwidth Manager: is responsible for wireless bandwidth allocation and management. Because the information needs to be forwarded to and by the neighbor stations, which will consume its bandwidth, the number of users of each base station needs to be rationally distributed to meet the disaster response demands regarding the number of communications channels of each base station to avoid allocation disequilibrium. If this task is not done well, the bandwidth of some base stations may be occupied by the transmitted information. Thus, 29.

(41) communication service cannot be provided. The worse-case scenario is that the bandwidth may be occupied by the less disastrous areas; the more disastrous areas may not receive any bandwidth at all. Given the optimal disaster response efficiency, it is necessary to allocate appropriate amount of communications channels to each base station.. Self-Adjustment: responsible for re-planning network topology and bandwidth allocation when the environment is changing.. 立. 政治大. ‧. ‧ 國. 學. 3.4. Deployment and Operation Procedures. The deployment of CCN Network is divided into four stages, with each stage elaborated. io. sit. y. Nat. below:. n. al. er. Stage 1: Damage Assessment Phase. Ch. i n U. v. The disaster response headquarter, which is most likely a government unit, will collect. engchi. disaster information and carry out a damage assessment to obtain an overall picture of the disaster. The CCN can be activated immediately to perform self-diagnosis if it is pre-installed in the existing cellular system. Before backup power is exhausted, an isolated station can self-diagnose the severity of the damage, identify the routing path to a survival station, and report the assessment to the control center.. Stage 2: Planning Phase: Choose the disaster areas and base stations for recovery. Design the recovery scheme according to the assessment, including network topology planning, 30.

(42) scheduling for the deployment of the topology, routing, and bandwidth allocation strategies, etc.. Stage 3: Deployment Phase: The construction and set-up of CCN are based on the results of the second stage.. 政治大. Stage 4: Operation Phase: The service strategies should be stipulated to allow ordered. 立. access to maximize the efficiency of disaster response operation. Priority based admission. 3.5. Basic CCN Services. Nat. y. ‧. ‧ 國. 學. control is one of the core functionality of this phase.. sit. Three communication services were designed to support disaster response: the ordinary. n. al. er. io. phone, walkie-talkie-like, and agency communication services.. Ch. engchi. i n U. v. Ordinary Phone Service: a regular telephone service similar to the Plain Old Telephone Service (POTS) with priority-based admission control.. Walkie-Talkie-Like Service: a push-to-talk, or walkie-talkie, group communication service. In a disaster response operation, people—especially trapped victims—may not know each other; thus, anonymous group communication (which does not require a caller to dial the 9- or 10-digit cell phone number of the recipient) is the most useful communication mode and constitutes the majority of CCN traffic. The CCN designates a 3-digit special number that 31.

(43) should fit the national dialing plan to enable users to turn their cell phones into walkie-talkies. This instruction can also be automatically sent to all cell phones within the CNN covered area by Short Message Service, thereby enabling even trapped victims to receive it. Notably, the scope of this service is confined to each base station to prevent the excessive consumption of resources.. Agency Communication Service: Each group of users with the same functional specialty can. 政治大. form an agency group with a dedicated, easy-to-remember phone number. Thus, every user. 立. can make a request to a specific functional agent by dialing the corresponding phone number.. ‧ 國. 學. Examples of agency groups are the headquarters, surgical doctors, blood suppliers, medical. ‧. suppliers, power cutters, and excavators. When a user makes a request by dialing the special phone number, the CCN rings a set number of members of the corresponding group near the. y. Nat. io. sit. caller. Any paged recipient can answer the phone, and the agency database can be preloaded. n. al. er. or registered in real time. However, the original phone number of a cell phone may be lost if a. Ch. i n U. v. CCN does not establish its connection to the core network; thus, the CCN must have its own. engchi. dialing plan and phone number designation process.. 3.6. Design Issues of the CCN During a real disaster, a CCN may proceed with a phased deployment: disaster assessment, planning, deployment, and operation. Each of these phases presents design concerns, and thus several critical planning and deployment challenges are discussed in this section.. 32.

(44) . Definition of Resource Profit Resource allocation and scheduling problems are usually modeled as combinatorial. optimization problems for which plenty of solutions are available. Some profit is earned when one unit of resource is allocated to a base station, where the objective is to maximize the total profit earned with limited resources. In a CCN design, profit should be measured by the efficiency with which it improves disaster response operations. Moreover, the definition of profit must be defined by a specific disaster response authority, because they possess the. 政治大. relevant real time disaster situation statistics as well as the priority determination authority. A. 立. typical profit definition when allocating a CRP to a base station is a linear or nonlinear. ‧ 國. 學. combination of the emergency level and the user population covered by the base station. The profit definition of allocating a communication channel to a pair of base stations is a major. ‧. challenge yet to be tackled. In the future, social scientists must formulate improved profit. y. Nat. n. al. er. io. . sit. definitions and optimization models.. Network Topology Design. Ch. engchi. i n U. v. According to the statistics we collected, more than 3300 base stations crashed during the 88 Flood, which was only a moderate scale disaster affecting a small number of counties in southern Taiwan. However, larger scale system crashes can be anticipated to occur in larger natural disasters. Therefore, the first concern when deploying a CCN is the selection of a limited number of crashed base stations based on the available CRPs as well as the topology for their connectivity; a simple tree-type or a resilient non-tree-type topology can then be formed with the objective of maximizing its efficiency and stability. In our study, the topology design [22,55] was formulated into different combinatorial optimization problems and solved using typical algorithmic methodologies [5]. To maintain the integrity of a CCN and adapt to 33.

(45) the constant changing conditions of a disaster, the forwarding topology must be replanted frequently. A system that has self-healing capabilities would be preferable, but requires further research. Tree topology is simple, but also vulnerable to a single link or node failure. In response, Charnrsriponyo [10,11] improved network reliability by maximizing the number of chains, and Elshqeirat [15,16] used a dynamic programming scheme to generate a topology comprising a selected sequence of spanning trees, thereby satisfying a predefined reliability.. 政治大. All nodes in the methodologies of Charnrsriponyo [10,11] and Elshqeirat [15,16] are identical.. 立. However, some pivot nodes such as command centers require higher bandwidth and reliability. ‧ 國. 學. than others. Hence, we proposed a better optimization model with differentiated reliability demands together with a Disjoint K-Path Max-Profit Mesh algorithm [22] to satisfy. ‧. bandwidth and reliability requirements by providing multiple path to the pivot nodes; thus,. y. Nat. n. al. er. io. . sit. every pivot node can reach the core network through two or more disjointed paths.. Deployment Scheduling. Ch. engchi. i n U. v. Some number of CRPs can be previously stored in the national disaster response center and be transported via some transportation vehicle such as helicopter to the selected stations to construct a CCN rapidly. The deployment sequence may have a big impact on the disaster response efficiency since the profit of restoring a base station keeps decreasing over time. The time variant survival rate, the difficulty of accessing, the delivery time, the topology constraints, and many other parameters all together makes the deployment scheduling a highly complex problem. Since the transportation capacity may be very limited, it may need several rounds of deployments. Unfortunately, the benefit of saving a station is gradually attenuated with time. 34.

(46) The deployment sequence will largely determine the disaster response efficiency. A good sequence may save more lives than a bad one. The transport sequence has not only to consider the emergency level of base stations but also to follow the scheduling constraint that ancestor nodes have to be recovered before their child. That’s because a child node cannot connect to the core network without the forwarding of its ancestors. CCN deployment scheduling problem is similar to fleet routing and scheduling problems [17] which aim to find an optimal route of one or more vehicles through a graph and assign. 政治大. vehicles to ideal routes at the particular time. Formulations of flee routing and scheduling. 立. problems are usually based on multi-commodity network flow problem or vehicle routing. ‧ 國. 學. problem. Objectives of these formulas are minimizing the unsatisfied demand or maximizing the demand satisfied. Formulas and solutions are proposed in [2,17,22]. Objectives of these. ‧. researches aim to minimize the transport time and the number of vehicles under the. y. Nat. sit. limitations of transportation capacity.. n. al. er. io. Researches of fleet routing and scheduling problems mainly consider the problem of how to. i n U. v. transport resource to disaster points with the shortest time and minimum cost. Beside. Ch. engchi. transport time and cost, there have more issues needed to be addressed in CCN deployment scheduling problem. First, the emergency levels of disaster areas are different. The profits of delivering materials to disaster areas to recover base stations are also different. Hence, the disaster area, which has higher emergency level, should have higher priority. Second, the profits are not constant but decreasing with time. Third, the deployment sequence has to follow the scheduling constraint. In order to solve these issues, a CCN deployment scheduling (CCN-DS) formulation is proposed [27]. And, a CCN-DS algorithm is used to find a heuristic deployment scheduling to approximately maximize the efficiency of disaster response operation. 35.

(47) . Priority Based Bandwidth Allocation and Admission Control The system capability of a CCN is mainly limited by the bandwidths of the external link. from the CCN to the core network, the inter-base-station links, and the radio channels from the base stations to cell phones. The bandwidth demand from users in a disaster typically far exceeds capacity. Therefore, some priority-based admission control must be implemented, in addition to a precise allocation plan that rationally distributes available bandwidth to base. 政治大. stations and maximizes CCN efficiency. Similar to the previous two problems, bandwidth. 立. allocation challenges were modeled into combinatorial optimization problems and solved. ‧ 國. 學. using heuristic algorithms [25].. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. 36. i n U. v.

(48) Chapter 4. Network Topology Design 4.1. Considerations of Network Topology Design In CCN, most disconnected base stations require multiple hops to connect to the core network. The network topology of CCN may have a great impact on the efficiency of CCN, which is the efficiency of disaster response operation and its stability. The objective of. 政治大 efficiency. The evaluation factors of disaster response efficiency include the emergency level 立. network topology design is to find a network topology to maximum the disaster response. ‧ 國. 學. of the afflicted areas or the level of the disaster and the number of disaster responders and victims. Referring to Fig. 4.1., the CCN network topology should deploy to those incident. ‧. areas that have large population or high emergency level to maximum the disaster response. sit. y. Nat. efficiency. However, some important areas, such as command centers, require higher. io. er. reliability and network bandwidth. These two requirements, maximization of the disaster. al. response efficiency and reliability, are dependent to each other. A good network topology. n. v i n design algorithm of CCN shouldC compromise two requirements. h e n gthese chi U. Besides, there are some constraints needed to be considered. The number of CRPs and the number of antennas in each CRP are fixed. Hence, the number of selected BSs is constrained by the number of CRPs; the number of selected links of each selected BS cannot exceed the number of antennas in each CRP. The packages are forwarded through multi-hop, too many hops may cause long delay time and consume too much intra-BS bandwidth. Therefore, the number of hops from some important areas to the core network should be limited to maintain appropriate quality of network service. Except these basic constraints mentioned above, CCN topology design need to be considerate 37.

(49) of more complicated problems. These problems are discussed as fellows.. Population. Candidate link. Command center. 立. 政治大. selected link. ‧ 國. 學. Isolated Selected. BS BS. ‧ er. io. sit. y. Nat. al. n. v i n C hConsiderations of U Figure 4.1. topoloy design engchi. Tree-type Topology vs. Mesh Network Topology Strengths of the tree-type network topology are easy to construct and maintain, and thus tree-type network topology is wildly applied. But, it is vulnerable to a single link or node failure. The mesh network topology increases the network availability by using multiple path to connect the core network and critical areas, such as command centers, such that every critical areas can reach the core network via two or more disjointed paths. However, it will have a side effect to the total profit when using multiple path. Hence, it needs to balance the network availability and total profit when employing multiple path. 38.

(50) Single Operator vs. Multiple Operators In general, the coverage areas of base stations are well planned and mutual exclusive. Therefore, it’s simpler to construct CCN network by using base stations from one operator than multiple operators. Nevertheless, enlarging the concurrent users of some critical areas may increase the disaster response efficiency. This can be achieved by including the base. 政治大 function of the base stations that cover the same area is a decreasing function of the number of 立 stations from multiple operators in the same covered area into CCN. However, the profit. base stations in service. The marginal benefit of profits will gradually decease when multiple. ‧ 國. 學. base stations covering the same area are selected into CCN.. ‧. Nat. sit. y. Depth Bound vs. Depth Weight. n. al. er. io. The base stations connect to the core network by multi-hops in CCN. In order to avoid long. i n U. v. hop connection and too much forwarding traffic, the number of hops has to be controlled.. Ch. engchi. Depth bound constrains the number of hops by setting an upper bound of the number of hops. This model is simple, but may discard a seriously damaged area that is too far from the root. Depth weight dynamically adjusts the tree depth so that the serious damage areas would not be discarded by the depth constraint.. 39.

(51) Simple FT. Cross FT. 政治大. 立. sit. y. ‧. ‧ 國. 學. io. n. al. Cross MPFN. er. Nat. MPFN. i n U. v. Figure 4.2. Four types of CCN network topology design problems. Ch. engchi. Combinations of these considerations discussed above are simple forwarding tree (Simple FT), cross-forwarding tree (Cross FT), multiple path forwarding network (MPFN) and Cross multiple path forwarding network (Cross MPFN), are proposed and discussed in this dissertation. Examples of these topologies are showed in Fig. 4.2. Simple FT design problem is to find a K-maximum spanning tree with degree bound. The original base stations of Simple FT belong to one operator. Dash lines denote the selected wireless links. Cross FT is also a tree-type topology. But, the original base stations of Cross FT are from multiple operators. Triangle and rectangle denote the base stations of different 40.