1. Centers for Disease Control and Prevention. Diabetes prevalence among American Indians and Alaska Native and the overall population—United States, 1994–2002. MMWR Morb Mortal Wkly Rep. 2003;52:702–704. [PubMed]
2. Burrows NR, Geiss LS, Engelgau MM, Acton KJ. Prevalence of diabetes among Native Americans and Alaska Natives, 1990–1997: an increasing burden. Diabetes Care. 2000;23: 1786–1790. [PubMed]
3. Acton KJ, Burrows NR, Moore K, et al. Trends in diabetes prevalence among American Indian and Alaska native children, adolescents, and young adults. Am J Public Health. 2002;92: 1485–1490. [PMC free article][PubMed]
4. Indian Health Service. Trends in Indian Health. Rockville, Md: Dept of Health and Human Services; 1999.
5. Edelman D, Olsen MK, Dudley TK, Harris AC, Oddone EZ. Quality of care for patients diagnosed with diabetes at screening. Diabetes Care. 2003;26: 367–371. [PubMed]
6. McGlynn EA, Asch SM, Adams J, et al. The quality of health care delivered to adults in the United States. N Engl J Med. 2003;348:2635–2645. [PubMed]
7. Chassin MR, Galvin RW. The urgent need to improve health care quality. Institute of Medicine National Roundtable on Health Care Quality. JAMA. 1998;280:1000–1005. [PubMed]
8. Agency for Healthcare Research and Quality. National Healthcare Disparities Report. Available at: http://qualitytools.ahrq.gov/disparitiesreport/download_report.aspx. Accessed December 20, 2004.
9. Institute of Medicine. Crossing the Quality Chasm. A New Health System for the 21st Century. Washington, DC: National Academies Press; 2001. [PubMed]
10. Bodenheimer T, Wagner EH, Grumbach K. Improving primary care for patients with chronic illness: the chronic care model, part 2. JAMA. 2002;288:1909–1914. [PubMed]
11. Hynes DM, Perrin RA, Rappaport S, Stevens JM, Demakis JG. Informatics resources to support health care quality improvement in the Veterans Health Administration. J Am Med Inform Assoc. 2004;11:344–350. [PMC free article][PubMed]
12. Jha AK, Perlin JB, Kizer KW, Dudley RA. Effect of the transformation of the Veterans Affairs Health Care System on the quality of care. N Engl J Med. 2003;348:2218–2227. [PubMed]
13. Dixon M, Roubideaux Y. Promises to Keep. Public Health Policy for American Indians & Alaska Natives in the 21st Century. Washington, DC: American Public Health Association; 2001.
14. Indian Health Service. Indian Health Service Year 2003 Profile. Available at: http://info.ihs.gov/Infrastructure/Infrastructure6.pdf. Accessed December 20, 2004.
15. Indian Health Service. Tribal self-determination. Available at: http://info.ihs.gov/TreatiesLaws/Treaties2.pdf. Accessed December 20, 2004.
16. Indian Health Service. Indians in urban areas. Available at: http://info.ihs.gov/People/People4.pdf. Accessed December 20, 2004.
17. Indian Health Service. Indian population. Available at: http://info.ihs.gov/People/People1.pdf. Accessed December 20, 2004.
18. Buchwald D, Sheffield J, Furman R, et al. Influenza and pneumococcal vaccination among Native American elders in a primary care practice. Arch Intern Med. 2000;160:1443–1448. [PubMed]
19. Buchwald D, Furman R, Ashton S, Manson S. Preventive care of older urban American Indians and Alaska natives in primary care. J Gen Intern Med. 2001;16:257–261. [PMC free article][PubMed]
20. Gilliland FD, Mahler R, Hunt WC, Davis SM. Preventive health care among rural American Indians in New Mexico. Prev Med. 1999;28:194–202. [PubMed]
21. Acton KJ, Shields R, Rith-Najarian S, et al. Applying the diabetes quality improvement project indicators in the Indian Health Service primary care setting. Diabetes Care. 2001;24:22–26. [PubMed]
22. Centers for Disease Control and Prevention. Health status of American Indians compared with other racial/ethnic minority populations—selected states, 2001–2002. MMWR Morb Mortal Wkly Rep. 2003;52:1148–1152. [PubMed]
23. Blackman DK, Bennett EM, Miller DS. Trends in self-reported use of mammograms (1989–1997) and Papanicolaou tests (1991–1997)—Behavioral Risk Factor Surveillance System. MMWR CDC Surveill Summ. 1999;48: 1–22. [PubMed]
24. Zuckerman S, Haley J, Roubideaux Y, Lillie-Blanton M. Health service access, use, and insurance coverage among American Indians/Alaska Natives and whites: what role does the Indian Health Service play? Am J Public Health. 2004; 94:53–59. [PMC free article][PubMed]
25. Virnig BA, Lurie N, Huang Z, et al. Racial variation in quality of care among Medicare+Choice enrollees. Health Aff. 2002;21:224–230. [PubMed]
26. Arday SL, Arday DR, Monroe S, Zhang J. HCFA’s racial and ethnic data: current accuracy and recent improvements. Health Care Financ Rev. 2000; 21:107–116. [PMC free article][PubMed]
27. Fleming BB, Greenfield S, Engelgau MM, et al. The Diabetes Quality Improvement Project: moving science into health policy to gain an edge on the diabetes epidemic. Diabetes Care. 2001;24:1815–1820. [PubMed]
28. Schneider EC, Riehl V, Courte-Wienecke S, Eddy DM, Sennett C. Enhancing performance measurement: NCQA’s road map for a health information framework. National Committee for Quality Assurance. JAMA. 1999;282: 1184–1190. [PubMed]
29. Nelson EC, Splaine ME, Batalden PB, Plume SK. Building measurement and data collection into medical practice. Ann Intern Med. 1998;128: 460–466. [PubMed]
30. James B. Information system concepts for quality measurement. Med Care. 2003;41(1 suppl):I71–I79. [PubMed]
31. Grol R. Improving the quality of medical care: building bridges among professional pride, payer profit, and patient satisfaction. JAMA. 2001;286: 2578–2585. [PubMed]
32. Ellrodt G, Cook DJ, Lee J, et al. Evidence-based disease management. JAMA. 1997;278:1687–1692. [PubMed]
33. Office of Management and Budget. Government Performance Results Act of 1993. Available at: http://www.whitehouse.gov/omb/mgmt-gpra/gplaw2m.html. Accessed December 20, 2004.
34. International Classification of Diseases, Ninth Revision. Geneva, Switzer-land: World Health Organization; 1980.
35. Jencks SF, Huff ED, Cuerdon T. Change in the quality of care delivered to Medicare beneficiaries, 1998–1999 to 2000–2001. JAMA. 2003;289: 305–312. [PubMed]
36. McDonald CJ, Huff SM, Suico JG, et al. LOINC, a universal standard for identifying laboratory observations: a 5-year update. Clin Chem. 2003;49: 624–633. [PubMed]
37. Fernandopulle R, Ferris T, Epstein A, et al. A research agenda for bridging the “quality chasm.” Health Aff. 2003; 22:178–190. [PubMed]
38. Choremis J, Chow DR. Use of telemedicine in screening for diabetic retinopathy. Can J Ophthalmol. 2003; 38:575–579. [PubMed]
39. Cavallerano AA, Cavallerano JD, Katalinic P, et al. Use of Joslin Vision Network digital-video nonmydriatic retinal imaging to assess diabetic retinopathy in a clinical program. Retina. 2003;23:215–223. [PubMed]
40. Davis RM, Fowler S, Bellis K, et al. Telemedicine improves eye examination rates in individuals with diabetes: a model for eye-care delivery in under-served communities. Diabetes Care. 2003;26:2476. [PubMed]
41. Liao Y, Tucker P, Okoro CA, et al. REACH 2010 Surveillance for Health Status in Minority Communities—United States, 2001–2002. MMWR Surveill Summ. 2004;53:1–36. [PubMed]
42. Hays H. The IHS Electronic Health Record Project. IHS Prim Care Provid. 2003;28:233–236.
43. Stone EG, Morton SC, Hulscher ME, et al. Interventions that increase use of adult immunization and cancer screening services: a meta-analysis. Ann Intern Med. 2002;136:641–651. [PubMed]
44. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects of computer-based clinical decision support systems on physician performance and patient outcomes: a systematic review. JAMA. 1998;280: 1339–1346. [PubMed]
45. Bates DW, Leape LL, Cullen DJ, et al. Effect of computerized physician order entry and a team intervention on prevention of serious medication errors. JAMA. 1998;280:1311–1316. [PubMed]
46. California Med-Cal Type 2 Diabetes Study Group. Closing the gap: effect of diabetes case management on glycemic control among low-income ethnic minority populations: the California Medi-Cal type 2 diabetes study. Diabetes Care. 2004;27:95–103. [PubMed]
47. Steinman MA, Gonzales R, Linder JA, Landefeld CS. Changing use of antibiotics in community-based outpatient practice, 1991–1999. Ann Intern Med. 2003;138:525–533. [PubMed]
48. Miller RH, Sim I. Physicians’ use of electronic medical records: barriers and solutions. Health Aff. 2004;23: 116–126. [PubMed]
49. Poon EG, Blumenthal D, Jaggi T, et al. Overcoming barriers to adopting and implementing computerized physician order entry systems in US hospitals. Health Aff. 2004;23:184–190. [PubMed]
50. Katz RJ. Addressing the health care needs of American Indians and Alaska Natives. Am J Public Health. 2004;94:13–14. [PMC free article][PubMed]
51. Wang SJ, Middleton B, Prosser LA, et al. A cost-benefit analysis of electronic medical records in primary care. Am J Med. 2003;114:397–403. [PubMed]
Wireless mesh network (WMN) is one of the wireless technologies to provide high-speed and cost effective wireless internet access service. WMN is self-configured, self-organized, and fault-tolerant to preserve the mesh connectivity after failure occurrence automatically. In fact, mesh nodes heal the failures by making instant and real-time decisions to find alternatives to transmit the data toward pre-defined destinations. All these features provide WMN with numerous overriding merits including flexibility, robustness, and reliability in service coverage, scalability to name a few in comparison with other similar technologies. These capabilities through relatively low initial cost, has attracted academia's attention and became increasingly popular in business, industrial, and personal applications in the recent years. In general, three types of mesh nodes play role in formation of WMN's architecture to provide the end user device with internet access: gateway, mesh router (MR) and mesh client (MC). Gateways serve the function of interconnecting the WMN MRs to the Internet. Wireless mesh backbone is formed by a number of stationary MRs. These intermediate nodes manage the network's traffics and are responsible for multi-hop connectivity provision among mesh hosts themselves and also between them and the Internet. Access points make such connectivity and as a result, make Internet access possible by forwarding MCs' traffics to MRs. The stationary or mobile MCs of different sub-networks such as Wireless Local Area Networks, worldwide interoperability for microwave access, or wireless sensor networks can connect themselves to the WMN's structure through the nearest MRs. Figure 1 represents the infrastructure of a given WMN.
Inter-channel interferences among neighboring nodes lead to dramatically capacity degradation and are of the most crucial challenges in traditional single-radio single-channel WMN. However, WMNs evolved to a new generation called single-radio multi-channels to exploit the advantage of using different non-interfering channels. However, the aforementioned problem had not been fully solved. Eventually, multiple radios have been proposed to switch on multiple channels to achieve higher throughput and to minimize channel conflict. By tuning the radio interfaces to the orthogonal channels, all interfaces can be utilized simultaneously with the minimum interference. Institute of Electrical and Electronics Engineers (IEEE) 802.11a standard operating on 5.8 GHz and IEEE 802.11b/g operating on 2.4 GHz presented 12 and 3 orthogonal frequency channels, respectively. Figure 2 depicts the available channels in IEEE 802.11 b/g.
Many communication mechanisms have been introduced in WMN including link scheduling [1, 2], gateway placement [3, 4], gateway selection [5, 6], relay selection , broadcast routing [8-11], and multicast routing [12-16] to support its applications. A vital question in all these designing issues is how to assign the available channels to interfaces of nodes in such a way that the minimal interference is achieved. Figure 3 shows an example of conducted channel assignment (CA) in a given MRMC-WMN.
Because the network performance heavily depends on CA efficiency, considerable numbers of dynamic and intelligent CA schemes have been proposed so far. CA algorithms have been designed to address various criteria such as interference, routing, delay, connectivity, link scheduling, stability, and congestion to name a few [17-19]. These methods can leverage other algorithms like graph theory [20, 21], game theory , and greedy approaches  to achieve the optimization goals. Generally, they can be categorized based on their main objective functions such as minimizing interference, maximizing throughput or congestion control [24-27]. Additionally, there is another classification including centralized [20, 22-25, 28-32] and decentralized [26, 33-42] approaches. In centralized ones, benefiting from comprehensive knowledge of the whole network, a central controller makes CA's decisions. The central node as the assigner computes the most suitable channel according to received request and then, informs its neighbors as well as the assignee of newly calculated channel. Software-defined networking is a promising solution to design a wireless network in which a central controller makes decisions to guarantee a stable performance. There are studies to improve the approach, for instance, in  wireless network virtualization is considered as complement to software-defined networking's management. On the other hands, there are no such central nodes in the decentralized methods; instead, each node can be both assigner and also assignee. It computes and selects its own channel independently and then announces its decision to others. As illustrated in Figure 4, in spite of diversity of CA schemes, they mostly include four main stages. In Figure 4, A, AE, and N depict the assigner, assignee, and neighboring nodes, respectively. CA algorithm is usually performed in two modes. Firstly, based on a pre-specified waiting time, the algorithm is run periodically (usually long time intervals). Secondly, it can be triggered after occurrence of significant changes in the network topology, traffic or channel loads. For instance, a light-loaded channel at the beginning can be a loaded and congested one after a while leading to interference. There are studies to tackle the interference, for example, in , collision probability is modeled, and in , a hopping approach changes channel to another light-loaded to tackle the interference of pre-allocation channels on slotted carrier sense multiple access with collision avoidance (CSMA/CA).
- Information gathering: It is the initial phase of a CA algorithm. In this phase, nodes listen on channels and collect the required data from their x-hop away neighbor nodes in case of decentralized or all nodes in centralized fashion. The collected data can be related to bandwidth, channel usage, hop-count distance from the wired gateway, measured load, throughput of interfering neighbors, the assigned superimposed codes, and link status to name a few [46-52]. The comprehensiveness and accuracy of data driven by this phase has direct impact on the node's ability to choose the best possible channel in the following decision-making step. Table I exemplifies the channel table of a given node in which, node 1 using channel 4 with measured load 100 is one-hop neighbor of a given node.
- Channel assessment: The given node computes the most proper channel based on the aggregated local data. It can consist of some sub-steps, for example, in , it is divided into three evaluation parts which will be explained in Section 3.1. Although, this computational phase differs in each proposed schemes, all have a common ultimate goal that channel c should be selected so that it leads to higher network performance.
- Channel notification: The neighboring nodes are notified of the selected channel to update their tables.
- Channel switching: Interfaces are tuned to the selected channels. In , assigner is required to switch its channel only after receiving assignee's approval as an acknowledgment message (ACK).
Regarding the importance of channel allocation, obviously selecting an improper channel is destructive and network throughput is affected seriously. It is crucial that all nodes play their defined roles in CA procedure so that their participations result in selection of best channel. However, in real networks, the trustworthiness of all participants including assigner, assignee and neighboring nodes are in doubt. Most of presented CA methods assumed implicitly that all nodes are well behaved, which is an ideal but impractical assumption. Indeed, their efforts could be easily wasted by a malicious mesh node. The presented study concentrates on the overlooked vulnerabilities in the majority of the CA algorithms causing some intrusions as well as the reported solutions against them in details.
The rest of this study is organized as follows. The existing attacks against CA algorithms are discussed in Section 2.1. Then, Section 3 surveys proposed anti-attack mechanisms. Finally, Section 4 concludes the paper.
2 Attacks and Vulnerabilities against CA Algorithms
There are some series of attacks on the CA protocols that are originated from four main vulnerabilities: (i) Almost all CA schemes trust irrationally to the mesh nodes that they behave well intentioned and present information as it is. This wrong assumption makes them easy to exploit. The most efficient and intelligent CA models which do not consider malicious intentions and have no defensive mechanism against them, will expose sever network performance degradation. Although, the majority of proposed schemes ignore the negative impact of malicious mesh nodes, recently, a few works have addressed the possible security drawbacks. In [46, 47], the possibilities of malicious assigner have discussed and in , some attacks were applied by the neighboring nodes. Additionally, the reliability of the assigner, the assignee, and the neighbor nodes has been challenged in [49-52]. (ii) There is no verification in the most CA procedure at all. Therefore, all kinds of nodes can violate the defined duties without any punishment or negative consequences. For instance, the fake information injected by the neighbors is used directly to accomplish CA computation improperly. Moreover, the lack of verification lets the malicious nodes choose an improper channel and notify others overtly. (iii) The probability of collusion among nodes have not been considered even in the suggested secure CA models. Most of them rely on a third trusted node verifying a received alarm message such as [46, 47, 49, 51]. Thus, none of them can detect a security sabotage while the nodes collude to not send any alarm. (iv) The attacks taking their roots from the aforementioned problems could not be prevented even by the basic security frameworks such as authentication and encryption [46, 51, 53]. For instance, security protocols using encryption provide an adversary node with a secure communication channel to send their misleading information. It is not possible to assess the information cognitively.
These weak points increase the overall channel interference leading to frequent channel re-assignments and re-connections. In the following, different kinds of attacks caused by misbehavior of malicious mesh nodes will be explained and analyzed in details.
2.1 Attacks by neighbors
To our best knowledge, there are some attacks launched directly or indirectly by malicious neighbors that will be argued in this section.
2.1.1 Forged information attack
Regarding its intention, the compromised node as neighboring node presents various fake, outdated, and unreal information. Figure 5 shows an example of the Forged information attack (FIA). The malicious node M lies about expected loads in channels 1 and 2 to newcomer nodes of G and D. Consequently, they choose these overloaded channels. This attack not only leads to improper channel decision, but it also deprives the victim node of regular data forwarding [48, 51-53]. Some schemes are designed to assign the channels after sensing considerable changes in measured values such as signal-to-interference-plus-noise ratio, traffic load, and link availability to name a few. CA algorithm can be intrigued frequently to stop its normal procedure and re-calculate for a new channel repeatedly by sending wrong environmental information.
2.1.2 Altering the neighbor information attack
An adversary may receive the information and forward its modification information to the neighbors and the channel assigners [51, 52].
2.1.3 Increasing the neighbor set attack
A malicious node increases the number of neighboring nodes by sending the base information on behalf of some phony nodes or even some theoretical nodes that are not neighbors at all owing to environmental factors [51, 52].
2.1.4 Decreasing the neighbor set or packet delivery attack
In this attack, an adversary as an intermediate node refuses to forward the data expected to be relayed toward an assigner with the aim of reducing the number of neighboring nodes [49, 51, 52].
2.1.5 Utilization-based conflict attack
Normally, an assigner is expected to select a channel having the largest non-occupied portion called channel margin . A malicious neighbor node intentionally manipulates the channels capacities to launch Utilization-based conflict attack (UCA). That is, it indirectly triggers the victim node to find another channel which causes less overall margin. Hence, further channel switching will be imposed on the victim. As shown in Figures 6 and 7, this attack can be applied in two scenarios: join UCA and leave UCA both through four a, b, c, and d steps. In the former, firstly, an adversary node M sniffs the aggregate channel margin of the target victim node V. Then, the node M uses the same channel (i.e., channel 2 in Figure 6b). Therefore, the node V observes a change in its local channel margin, and it runs its CA algorithm to find a channel with a higher capacity (i.e., channel 1 in Figure 6d). The UCA indirectly imposes a channel switching overhead on the victim node to keep it busy and suspend its data transmission.
In the latter attack, in order to influence the channel margin of the victim node V, the malicious node M switches its own channel to another (i.e., channel 3 from 1in Figure 7b). Therefore, the node M changes observed channel margin of the node V making it pause its data transmission (i.e., with node A) and get ready to run its CA algorithm. This attack would succeed if the victim node switches to another channel with less channel margin (i.e., channel 4 compared with channel 2 in Figure 7d). As it will be discussed later at Sections 2.1.6 and 2.2.5, this attack is part of other attacks named LIBA and Denial of data attack .
2.1.6 Link break attack
Once a node changes its channel, it notifies its neighbors about the recent channel switching. The connection would be lost unless they update their channel table in a timely manner. This point is exploited by two attacks: direct Link break attack (LIBA), indirect LIBA. As it is depicted in Figure 8a, in the former attack, the malicious node M triggers node A to change its channel. Then, node A aims to inform node B to change the communication channel. This notification does not get to B because of occurred interference with M. Consequently, the victim nodes of A and B cannot agree on the new selected channel. As shown in Figure 9b, in the latter attack, a malicious node M makes two victim nodes B and C change their channels simultaneously (i.e., by UCA) with the object of breaking their link. In this scenario, because both end nodes change their own channels, they cannot hear the incoming notification messages .
2.1.7 Security alarm attack
In some CA models like [47, 52], neighboring nodes receiving abnormal and conflicting data send a message called security alarm message (SAM). Because of lack of verification, it can be exploited to damage a victim's reputation and introduce it as a suspicious node. As an instant scenario, the node N gets ready to switch its own channel relieving the heavy traffic. The malicious node M broadcasts the fake SAMs pretending the misbehavior of node N. Because the SAMs are not verified, the gateway will assume the received requests or messages from N are invalid, and the channel cannot change.
2.1.8 Radio jamming attack
This attack makes a channel disable by sending continuous or periodic network traffics or noise signals. It targets on disrupting the normal operation of the network and dropping its capacity [48, 54].
2.2 Attacks by assigners and assignees
In addition to the discussed attacks by neighbors, there are some attacks launched by malicious assigners and assignees with the aim of network performance degradation that will be argued in this section.
2.2.1 Inappropriate channel assignment attack
Under the normal CA operation, an assigner requires to assign the most efficient and the least loaded possible channel to its interfaces. However, a malevolent assigner deliberately selects an improper one. For instance, the channel can be chosen causing the highest interference [46, 49, 51, 52] (e.g., allocating channel 1 by adversary node M in Figure 9a) or at least not causing the least possible interference . Additionally, a malicious assigner assigns different channels to assignees intending to communicate with each other through a common channel. For example, channels 1 and 2 are assigned to nodes B and A in Figure 9b, respectively. Furthermore, the Inappropriate channel assignment attack (ICAA) can also be initiated once an assigner limits its choices to only a part of available non-overlapped channels [51, 52]
Network Endo Parasite Attack (NEPA)  is another ICAA modeled in hyacinth algorithm . The assigner not only chooses a high priority channel used by wired gateways, but it also does not notify other nodes other than its own children. It keeps the network unaware of the changes to increase the hidden usage of heavily loaded channel and finally aggravates interference in communication links over these channels. Figure 10 shows an example of the NEPA. Here, node M which is assumed to be a compromised node has assigned the channels on links MH and MI as the same channels used by links CG1 and DG1, respectively, which will raise the interference on links CG1Cand DG1. Moreover, M does not inform its neighbors about its decisions.
Similarly, in Channel Ecto Parasite Attack (CEPA) , the attacker assigns all its interfaces to the highest priority channel without any notifications to neighboring nodes. This attack is shown in Figure 11, where node M as an attacker switches channels used by MH and MI links to the channel of CG1 link as the heaviest load channel. Because the end points of C and G1 trust all nodes, they consider the high interference is the result of an external factor like noise.
2.2.2 Forging notification attack
Notification phase can also be exploited in two ways [46, 47, 49, 51, 52]: First, a malicious node switches its interfaces and like the NEPA and the CEPA skips the notification step. Second, like Low-cost Ripple effect Attack , an adversary sends the forged and wrong channel number as a new assigned channel in order to tear down the links and increase the interference. The malicious node pretends to change its channel while it has not been changed in deed. In fact, it only sends misleading channel information as a new assigned channel. Because in this attack there is no need to actually change the channel, it is a low cost one. Figure 12 illustrates the Low-cost Ripple effect Attack. The compromised node M deceitfully informs its neighbors telling them that links MF and MH have been switched to new channels, although they are not changed. Based on these forged notifications, the neighbors change their assigned channel to links IJ, DE and, BD, which may change channels of A and C nodes too.
2.2.3 Improper switch attack
An adversary assigner or assignee switches its channel to something different from notified or assigned channel [51, 52].
2.2.4 Spam request attack
In centralized CA approaches, an attacker assignee node requests a free channel, while actually it does not need it, at all . Also, in decentralized CAs, each node as assigner can choose a channel and use it to transmit some unnecessary information with the aim of occupying it and preventing other nodes to access it.
2.2.5 Denial of data attack
As shown in Figure 13, in this attack, a malicious node makes a victim node in normal-mode to go back to the information gathering or CA computation phase . There are two ways for a malicious node to make this happen; DoD transition and catch-and-release DoD strategies. In the former strategy, the attacker intentionally tears down the connections; for example, an assignee or assigner leaves or changes its channel to disturb normal packet transmission. In the latter, an assignee or assigner causes the intentional failure through two steps; firstly, the LIBA is launched to break the victim's connections to persuade it to use attacker's channel. Afterwards, the malicious node does not transfer any data to be in silent mode. The objective of this approach is that a victim experiences link failure when it sends data packets to an attacker.
Finally, all the attacks discussed earlier may lead to ripple effect attack (REA). For example, this issue may occur once channel switching of a node leads to more switching across the whole network. Consequently, the network will slow down because all neighbors have to suspend their normal data packets transmissions responding to the recent changes. Therefore, the whole network would be tightly influenced by the REA in comparison with the RJA that only affects the attacker's neighborhood . The Figure 12 illustrates the effect of the attack by the arrows.
As it is being discussed, there are some attacks on CA algorithm in MRMC-WMNs. Figure 14 shows a chart briefing of the presented attacks. However, a verifier node that has been proposed recently as a defensive mechanism (explained in Section 3.1) is assumed reliable and attacks related to malicious verifier are overlooked so far. Moreover, there are some more complex attacks in which mesh nodes unify against a victim node, such as collusion of the assigner and the assignee or the neighboring nodes and the verifier to name a few. Table II compares security based studies on CA algorithm in which different nodes are considered to be malicious.
3 Anti-Attack Strategies versus Channel Assignment Mechanisms
As it was explained in Section 2, CA algorithms suffer from lack of verification and mesh nodes can be compromised based on their roles. Thus, most of the reported solutions have augmented a kind of verification phase to monitor the behavior of mesh nodes with the aim of controlling information accuracy and independent decisions. These verification mechanisms will be explained in the two following sections.
3.1 Solutions against malicious neighbors
3.1.1 Forged information attack defensive mechanism
The proposed CA algorithm in  applies two steps against FIA. Initially, two-hop mesh nodes exchange the information of their channels or expected load on their links. Afterwards, the receiver node compares the gathered neighbors' information. Inconsistent information triggers the receiver node to ask a trusted node about the suspected neighbors. The trusted node is selected based on the highest average degree of trust value calculated from packet drop, packet loss, and CA attacks criteria. The reference  also presents another comparison based protocol. Firstly, the received messages are verified by one-hop neighbors. Second, when no anomaly is found in the verification phase, a positive control message would be sent to notify the message is credible. Otherwise, a SAM is sent to notify its abnormality and a channel scan is requested. A low-load node takes charge of scanning which is randomly selected by a randomized algorithm called scanner selection strategy. Once the message is again confirmed to be illegal, a SAM is sent to notify that it should be discarded. The reference  proposed protocol in which a three-step process (channel evaluation, channel switch cost evaluation, and randomized selection) takes place before each channel tuning. In the channel evaluation phase, the assigner computes an Eval value of channels using Eq. (1), where C1, C2, W1, and W2 are numbers and associated weights of one-hop and two-hop neighboring mesh nodes, respectively. This study considers W2 less than W1 (e. g. W1 = 0.7, W2 = 0.3) to make the protocol less sensitive to environmental changes caused by forged two-hop nodes' information. In channel switch cost evaluation, the node compares the Eval with pre-defined threshold to decide whether or not a channel conflict is unbearable and therefore, a CA initiation is vital. In this way, it is claimed that the system does not react to slight changes immediately, and it is designed to tolerate and switch its channel only if a new channel is better significantly. If channel changing is inevitable based on the previous step, the node decision of alternative channel is not predictable because it is chosen randomly among a preferred non-busy channels list formed based on the Eval values. It is believed that the randomized selection strategy alleviates the slow REA too.