A Novel Congestion-Aware Interest Flooding Attacks Detection Mechanism in Named Data Networking Ahmed Benmoussa∗ , Abdou el Karim Tahari∗ , Nasreddine Lagraa∗ , Abderrahmane Lakas† Farhan Ahmad‡ , Rasheed Hussain§ , Chaker Abdelaziz Kerrache∗¶ , Fatih Kurugollu‡ ∗ Department of Computer Science, University of Laghouat, Algeria College of Information Technology, United Arab Emirates University, Al Ain, UAE ‡ Cyber Security Research Group, College of Engineering and Technology, University of Derby, Derby, UK § Institute of Security and Cyber-Physical Systems, Innopolis University, Innopolis, Russia ¶ Department of Maths and Computer Science, University of Ghardaia, Ghardaia, Algeria Emails: ∗ {ah.benmoussa, k.tahari, n.lagraa, ch.kerrache}@lagh-univ.dz † alakas@uaeu.ac.ae ‡ {f.ahmad, f.kurugollu}@derby.ac.uk § r.hussain@innopolis.ru † Abstract—Named Data Networking (NDN) is a promising candidate for future internet architecture. It is one of the implementations of the Information-Centric Networking (ICN) architectures where the focus is on the data rather than the owner of the data. While the data security is assured by definition, these networks are susceptible of various Denial of Service (DoS) attacks, mainly Interest Flooding Attacks (IFA). IFAs overwhelm an NDN router with a huge amount of interests (Data requests). Various solutions have been proposed in the literature to mitigate IFAs; however; these solutions do not make a difference between intentional and unintentional misbehavior due to the network congestion. In this paper, we propose a novel congestion-aware IFA detection and mitigation solution. We performed extensive simulations and the results clearly depict the efficiency of our proposal in detecting truly occurring IFA attacks. Keywords—Named Data Networking, Interest Flooding Attack, Denial of Service, Network congestion. I. I NTRODUCTION According to Ciscos Visual Networking Index, the Internet traffic will reach 254 Exabytes per month by 2020. With the constantly growing number of connected objects (nearly 20 billion in 2018) and the sheer amount of data exchanged, a new and more suitable Internet architecture is needed. The need for new Internet architecture is because of the change in the perspective of Internet usage. The Internet users are more interested in the content rather than in the sources of these contents. Furthermore, the content should reach the consumers as soon as possible [1]. Besides that, security was an afterthought when the actual Internet architecture was developed. Mobility and multicast were also adapted years after the Internet was created and they are still not fully adapted with the actual Internet [2], [3]. With these new challenges come new solutions. One of the proposed future Internet architecture projects is Internet Centric Networking (ICN). Several architectures were proposed under the ICN project. Named Data Networking is the promising instantiation of the ICN. NDN changes the network paradigm from host-centric to data-centric [4]. NDN incorporates intrinsic security to Internet by providing security to the contents rather than to the communication path. However, it is still not immune against attacks like DoS/DDoS. One of the DoS/DDoS attacks that targets an NDN node is Interest Flooding Attack (IFA). The goal of the attacker is to overwhelm the NDN node with Interests in order to make it unavailable. Several solutions were proposed to mitigate the IFA. Some solutions are local and others collaborative, distributed, and/or centralized [5]–[11]. These solutions use network traffic statistics to detect IFAs; however, the type of statistics they use, might be different. The existing solutions do not take network congestion into account. In fact, no existing solution take the network congestion as a parameter for detecting IFA which can lead to false alarms. Besides the conventional parameters, in this work, we consider network congestion as an additional parameter for detecting the IFA. Hence, our proposal can detect whether there is an intentional IFA or unintentional phenomenon caused by the congested network. The rest of this paper is organized as follows: In section II, we briefly describe the NDN architecture, the forwarding process and finish this section by talking about IFA. In section III, we present our solution to detect and mitigate an IFA. Afterwards, in Section IV, we evaluate our proposed solution. Finally, in section V, we conclude the paper. II. NAMED DATA N ETWORKING : A B IRD ’ S V IEW Named Data Networking (NDN) is a candidate architecture for Future Internet Architecture (FIA) to replace traditional TCP/IP Internet architecture [4], [12]. NDN is an extension of an earlier project in the domain of FIA named ContentCentric Networking (CCN). NDN architecture relies on the data-centric paradigm where data is the pinnacle of the communication architecture rather than the hosts. NDN uses data names instead of IP addresses for communication, 978-1-7281-1856-7/19/$31.00 ©2019 IEEE Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on December 08,2020 at 03:30:32 UTC from IEEE Xplore. Restrictions apply. which are hierarchically structured and meaningful, e.g., this paper may have the name: "org/icccn/28/ndn/A Novel-Congestion-Aware-Interest-FloodingAttacks-Detection-Mechanism-in-Named-DataNetworking" NDNRouter InterestPacket ContentStore (CS) PendingInterest Table(PIT) AddIncoming interface ReturnData Forwarding InformationBase (FIB) Forward Interest Dropor NACK A. NDN Architecture NDN uses two types of packets for communication among nodes, i.e., (1) Interest packets and (2) Data packets. Interest packets are used by end users to request specific data, normally known as ”consumers” in NDN. These interest packets propagates through the NDN network in search of the data requested by the consumers. Any node, having this data responds back to the consumer with a data packet. These data producing nodes are known as ”producers” in NDN. To fetch specific data, NDN node sends an interest with the name of the desired data. The network then tries to find this data from the closest possible location. Further, every data packet in NDN is digitally signed to keep the data intact. In order to disseminate interest and data packets in the network, every NDN node is equipped with following three data structures, i.e., • Pending Interest Table (PIT): PIT corresponds to the repository where the actual, not yet satisfied interest packets are stored. These interests will be held until they are satisfied or timed out. Every PIT entry contains the name of the requested data and the interfaces from where these interests came from. • Forwarding Information Base (FIB): FIB represents the second data structure of NDN core, which acts in a similar fashion as IP routing table. However, in NDN, FIB stores the namespaces instead of IP addresses. • Content Store (CS): CS within the NDN node acts as a cache node and is used mainly for caching purposes. CS integrates various cashing policies such as least recently used (LRU), first in first out (FIFO), to name a few. Every node in NDN can store the data that passes through it depending on the node’s caching policy. B. NDN Forwarding When an interest arrives at a router, it first checks the existence of the requested data in the CS. If a copy of the requested data is already stored, the router will then sends back this data packet downstream. If the CS does not have the requested data, the router checks if there is a pending entry of this interest in the PIT table (i.e., the router received a request for this data before). If the PIT table already contains an entry for this data, it will add to this entry the interface from where this interest came from. If there is no entry in the PIT table (e.g., there is no pending request for this data), the router will add a new entry to the PIT table with the name of this requested data. Then, the router uses the FIB table to route the interest packet to the next node (router or producer). When an interest is satisfied, a data packet is sent back. When a router receives a data packet, it will first check if this data was requested before by checking the respective entry ReturnData CacheData PendingInterest Table(PIT) ContentStore (CS) DataPacket (Signed) DropData Fig. 1. Request/Response operations in NDN in the PIT table. If there is no entry in the PIT table, the router will discard this data packet. If there is an entry in the PIT table, the router will choose to store (or not) this data packet (depends on the caching strategy) before forwarding this data downstream. The process of sending/receiving packets in NDN is described in figure 1. C. Interest Flooding Attack Interest Flooding Attack (IFA) is a type of attack that targets NDN nodes directly as depicted in figure 2. It consists of overwhelming an NDN router with a huge amount of interests in order to charge the PIT table, so there will be no place for the incoming legitimate interests. The initiator of an IFA (could be one or many, local or distributed) sends a large number of interests to target one or multiple routers within the network. The attacker can perform IFA in three major ways: 1) Sending a large number of interests requesting existing data. This scenario is not durable because of the routers’ capacity to store the data. Therefore, when the data is requested, it will be quickly satisfied and the attack will no longer take effect, unless the attacker wants to target the router with a cache pollution attack which is not in the scope of this paper. 2) By requesting dynamic data. In this scenario, the attacker floods the victim with interests requesting dynamic data (which are created dynamically after a request). It is more harmful than the first scenario because the data is created dynamically, so every interest should go to the real producer of this data. Due to this reason, caching data in the intermediate routers will not help in this scenario. 3) Requesting non-existing data. This is the most dangerous and the most used type of requests for performing an IFA. In this scenario, the initiator sends a large number of interests with invalid names. The goal of the attacker is to make the interests packets stay in the PIT until they time out because these requests will never be satisfied. The third scenario is the hardest to mitigate among them. Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on December 08,2020 at 03:30:32 UTC from IEEE Xplore. Restrictions apply. Discarded Interest NDN Router 315 Ignored User Upth = 310 eth1 Number of Interest packets 310 Full PIT Router A Producers CS eth2 Attacker NDN Network Legitimate Interests Forged Interests 305 10% ; α = 0,00001 Avg = 300 300 295 290 285 50% ; α = 0,1 Lowth = 150 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162 169 176 183 190 197 204 211 218 225 232 239 246 253 260 267 274 281 288 295 280 Seconds Fig. 2. Exemple of Interest Flooding Attack III. P ROPOSED IFA D ETECTION M ECHANISMS IN NDN Fig. 3. Average rate calculation example A. Satisfaction Ratio The first parameter used in our solution is the satisfaction ratio. It represents the number of received data packets by the number of issued interest. Satisf action Ratioi = # of Data P ackets Receivedi # of Interest P ackets Issuedi (1) Every router monitors its interfaces and calculates the satisfaction ratios, periodically, associated with these interfaces. This parameter will then be used by routers to detect malicious interfaces. When the satisfaction ratio of an interface is low, the router will consider this traffic as suspicious. The satisfaction ratio could lead to false detection of malicious interface (e.g., when the network is congested or the data producer is not reachable). Therefore, we induced network congestion and interest rate parameter within our proposed solution to get reliable detection results. B. Incoming Interest Rate This parameter represents the amount of interest packets arriving at the specific interface of NDN router in a given amount of time. This parameter is used by the routers to detect abnormal rates, where, every router monitors the incoming rate of each interface. When the rate exceeds the average threshold level, the routers will consider it as suspicious and starts investigations about the interest packets. The router calculates the average rate of each interface independently, which is obtained as follow: N ewAvgi = (1 − α)OldAvgi + α × inrit (2) In equation 2, N ewAvgi is the newly calculated average of interface (i), OldAvgi represents the actual average associated with interface (i), while inrit is the incoming rate to interface (i) of the router at time (t). Furthermore, the used skew factor (α) is calculated as: 0.1, if inr ∈ [Lowth , Avgi ] α= 0.00001, if inr ∈ [Avgi , U pth ] where, Avgi represents the average rate of interface (i). U pth is the upper bound, which is given as avg + 10%Avg while Lowth represents the lower bound threshold with value of avg − 50%Avg. In equation 2, we chose these values of α after a series of tests. The main motivation of these tests was to identify and chose a value that doesn’t influence on the average when an IFA is happening. We also decided to take in consideration only the 10% (above average) part. So if an attack is happening, it will not make the average value increases. For the lower bound α, we decided to take only the 50% below average. We decided to exclude the other part so it will not make the average rate decreases very quickly (e.g., when interface is not active). The figure 3 shows an example of rate average and upper/lower thresholds. C. Network Congestion The third and last parameter in our proposed solution is network congestion. The network congestion could lead to false results upon detection of an IFA. When a router adds an interest to the PIT table, it assigns with this interest a lifetime (default 4 seconds). This newly added interest will stay in the PIT table as long as it is not yet satisfied (data packet returned) or timed-out (no data packet returned). When the network is congested, interest and data packets take a long time to reach destination. This delay makes the pending interests in the PIT table to timeout, i.e., the pending interests times out before they are satisfied. In this paper we chose a simple way to detect a network congestion. The detection process is done locally by every router. In our solution, we chose two parameters to detect a network congestion: • Number of timed-out interests: when a network congestion occurs, the number of timed out interest packets increases. This is due to the delay as the pending interests take longer time to be satisfied. • Number of NACK packets: NACK packets are used to carry the response to an interest packet which is not satisfied. When the network is congested, the number of NACK packets arriving in a period of time decreases because of the delay. We compare these two parameters to get the state of the network (congested or not). If the numbers are not identical or nearly identical, this means that the network is congested i.e., the interest packets are timed out and no NACK packet Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on December 08,2020 at 03:30:32 UTC from IEEE Xplore. Restrictions apply. returned. Another parameter that could be used to detect a network congestion in NDN is the PIT table occupation. When a network is congested, interest packets take longer time in order to be satisfied, thus, staying longer in PIT tables of NDN routers. D. Proposed Algorithm The routers monitor the incoming rate of their interfaces. The average is calculated periodically by calling the Average procedure. When an incoming traffic rate of interface (i) exceeds average rate of of this interface, the router will consider it as suspicious and starts to investigate. The router will first check the satisfaction ratio of this interface (by calling the Procedure SatisfactionRatio). If the satisfaction ratio is low, the router will check the network state. The router will invoke the Procedure IsCongested to test if the network is congested or not. When all parameters are satisfied, i.e., incoming rate is below average, the satisfaction ratio is low and the network is not congested, The router will consider this interface as malicious and block it. The algorithms 1, 2, 3 and 4 show these procedures in details. Algorithm 1 Rate average calulation 1: procedure AVERAGE(int OldAvg, int inr) 2: Avg ← OldAvg 3: if ((Avg − (50%Avg)) ≤ inr ≤ Avg) then 4: Avg ← 0.9 × Avg + 0.1 × inr 5: else if (Avg ≤ inr ≤ (Avg + 10%Avg)) then 6: Avg ← 0.99999 × Avg + 0.00001 × inr 7: end if 8: return Avg 9: end procedure Algorithm 2 Satisfaction ratio calculation 1: procedure S ATISFACTION R ATIO(interface i, time t) 2: float ratio ← 0 3: Integer int nbr ← Number of Interest packets sent by interface (i) in periode t Integer data nbr ← Number of Data packets received by interface (i) in periode t 5: ratio ← data nbr/int nbr 6: return ratio 7: end procedure 4: Algorithm 3 Testing Network Congestion 1: procedure I S C ONGESTED 2: bool congested ← f alse 3: Integer timeoint nbr ← Number of Timed-out interests sent 4: 5: 6: 7: 8: 9: by all interfaces Integer nack nbr ← Number of NACK received by all interfaces if (timeoint nbr >> nack nbr) then # of timed-out interest is much bigger than # of Nack congested ← true end if return congested end procedure Algorithm 4 Detecting an Attacker 1: procedure I S ATTACKER(interface i, int inr) 2: bool attacker ← f alse 3: int Avg ← GetAvg(interf ace i) Returns average rate of interface (i) int ratio ← Satisf actionRatio(interf ace i) bool congested ← IsCongested if (inr > Avg) AND (ratio is low) AND (congested = false) then 7: Block the interface (i) 8: end if 9: end procedure 4: 5: 6: IV. P ERFORMANCE EVALUATION In this section, we provide details of the simulation settings and the simulation results of our proposed solution to mitigate IFA. A. Simulation Settings We evaluated our proposal using ndnSIM, which is an open source simulator to design and model NDN networks [13]. In order to simulate our proposal to detect and mitigate IFA in NDN, we considered a topology with seven routers, nine producers and nine consumers. The attacker model used in our simulation continuously sends huge amount of forged interest packets with constanct bit-rate, i.e., (10000 interests/s). Further, the consumers request data as follows: ‘Consumer1’ requests data from ‘Producer9’ using namespace "/dest9/", ‘Consumer2’ requests data from ‘Producer8’ using namespace "/dest8/" and, so on. Simulation details are provided in Table I. In order to evaluate IFA, we start our simulations by showing the effects of a congested network on the overall performance. Afterwards, we show that the network congestion leads to false detection of IFA. In addition, we prove that existing solutions relying on interest satisfaction ratio are not effective when a network is congested. Finally, we simulate an attack and show how the network is affected and how our proposed solution solves the false detection decisions due to the congested network. B. Normal scenario We start by simulating a normal scenario in order to compare the results with other scenarios. The results show that for a normal scenario, satisfaction ratio of nearly 100% is achieved for every consumer. That means that nearly every interest packet sent by consumers are satisfied (data packet returned). In the next section, we will provide comparison of these results with a congested network. C. Congestion scenario In this scenario, we simulate a congested network to see the resultant impact on the NDN network. To simulate a congested network, we restricted the throughput of the link between Router1 and Router7 to maximum 1 Mbps and introduced the delay of 100ms in this particular link. The simulation results for this scenario are given as: Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on December 08,2020 at 03:30:32 UTC from IEEE Xplore. Restrictions apply. TABLE I S IMULATION SETTINGS Parameter Value Simulation time Number of nodes Data Packet size CS size Cache replacement strategy Forwarding Strategy Average rate Rate of legitimate users Rate of attackers Number of Attackers 300s, 150s 25 1,024Bytes 100 LRU Best Route 300 interests/s 100 to 500 interests/s 10000 interests/s 01 700 Link Throughput Delay Router Router Router Producer Router Consumer Congested link 100Mbps 10Mbps 10ms 10ms <1Mbps 100ms # of interests 600 500 400 300 200 100 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101 106 111 116 121 126 0 Time (s) Fig. 5. Number of Dropped packets by Router7 (Core Router) 14% 12% 10% 600 8% 500 6% 400 4% 300 2% 200 0% 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101 106 111 116 121 126 131 # of timed-out interests 700 100 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101 106 111 116 121 126 131 136 0 Consumer1 Consumer2 Consumer3 Time (s) Fig. 6. Satisfaction ratios of Consumers 1, 2 and 3 Fig. 4. Number of timed-out Interests of Router1 1) Timed-out Interest Packets: We start by showing how the network congestion impacts the timed-out interest packets. The results shows that more than 90% of issued interest packets are timed-out for congested network, which is a very huge number. Further, the effect of network congestion on the PIT of Router1 is depicted in figure 4, suggesting that about 90% of pending interests are timed-out. This is due to the reason that the network congestion in the network results in the delay of the interest packets and as a result, the network experience to drop these interest packets. 2) Dropped Packets: In this section, we show how the network congestion impacted Router7 (Core router). The figure 5 shows the number of dropped packets by the core router. It is a considerable number when compared to the total number of packets passing through the Core Router (1600 packets/s). 40% of packets are dropped. All the network packets (interest and data packets) are passing through the core router. 40% of whole network traffic is dropped. 3) Satisfaction Ratio: In this section, we analyze the impacts of a congested network on the satisfaction ratio. In this section we prove that the existing solutions may give false detection of an IFA. We can see that a congested network can significantly decreases the satisfaction ratios of router’s interfaces. For example Consumer1 who requests 100 interest/s is receiving an average of only 5 Data packets/s which means a satisfaction ratio of 5%. The figure 6 shows the satisfaction ratios values associated with consumers 1,2 and 3. Solutions that relies only on the satisfaction ratio (e.g., Token Bucket [6]) are inefficient when a network is congested. In this section we showed how a congested network can impacts the statistics of a router and may lead to mistakenly detect an IFA. The other nodes in the network were not affected by the network congestion. D. Attack scenario The last simulation scenario is the IFA scenario. In this scenario the Consumer3 is malicious. He is requesting 10000 of forged interest packets per second. He is using the namespace "/dest7/attack" to request data from Producer7 who satisfies interest packets in the namespace "/dest7/legit". The goal of the attacker (Consumer3 in this scenario) is that the interests arrive to destination (Producer7) without being satisfied so they will stay in PIT tables until they time-out. In this scenario we see a huge number of NACK packets. This is the result of non existing data. The producers when they receive an interest for non existing data they responds Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on December 08,2020 at 03:30:32 UTC from IEEE Xplore. Restrictions apply. 12000 10000 10000 8000 8000 6000 # of packets # of packets 12000 Attack period 4000 2000 6000 Attack period 4000 2000 0 0 0 5 10 15 20 25 30 35 0 5 10 15 Time (s) Attack started 20 25 30 35 Time (s) Attack blocked Attack started Attack blocked Fig. 7. Number of NACK packets receiver by Router1 Fig. 8. Timed-out interest packets by Router1 with NACK packets. The figure 7 shows the number of NACK packets received by Router1 (Attacker’s Gateway). The number of NACK packets starts to rise when the attack is launched (at 10th second). The number of received NACK packets will be stabilized at the value of 10000/s which equals to the number of interest packets sent by the attacker. But when the attacker is mitigated (blocked), the number of NACK packets goes down to 0 NACK packet/s after 3 seconds. Another parameter that changes when an attack is happening is the timed-out interests. The figure 8 shows the number of timed-out interest packets by Router1. We also noted some timed-out Interest packets of legitimate users. This is because of the attack but the number is not considerable compared to the timed-out interest packets sent by the attacker. The number of Timed-out interest packets in Router1 goes up in t = 10s second (attack launch) until it stabilizes under 10000 timedout interest packets per second. Then this number starts to go down in t = 30s when the attacker is blocked. This number will goes down to 0 packet/s after 3 seconds. In the results of figures 7 and 8 we see that the number of NACK packets and timed-out interest packets are identical. For our algorithm this mean that the network is not congested. The not congested network is used by our algorithm to detect an IFA initiator. R EFERENCES V. C ONCLUSION AND F UTURE D IRECTIONS In this paper, we proposed a solution to detect and mitigate the Interest Flooding Attacks in NDN. Our proposal takes into consideration the network congestion and take it as a parameter to avoid false alarms regarding the behavior of the routers. Through simulation results we showed that the network congestion can lead routers to mistakenly consider a legitimate user as malicious. Finally we presented a scenario of an IFA and showed how the attacker was mitigated. As future works, we plan to extend our solution to detect other attacker models. We are also planning to develop a collaborative network congestion detection. [1] J. Rexford and C. Dovrolis, “Future internet architecture: clean-slate versus evolutionary research,” Communications of the ACM, vol. 53, no. 9, pp. 36–40, 2010. [2] R. Tourani, S. Misra, T. Mick, and G. Panwar, “Security, privacy, and access control in information-centric networking: A survey,” IEEE communications surveys & tutorials, vol. 20, no. 1, pp. 566–600, 2018. [3] E. G. AbdAllah, H. S. Hassanein, and M. Zulkernine, “A survey of security attacks in information-centric networking,” IEEE Communications Surveys & Tutorials, vol. 17, no. 3, pp. 1441–1454, 2015. [4] L. Zhang, D. Estrin, J. Burke, V. Jacobson, J. D. Thornton, D. K. Smetters, B. Zhang, G. Tsudik, D. Massey, C. Papadopoulos et al., “Named data networking (ndn) project,” NDN-0001, Xerox Palo Alto Research Center-PARC, vol. 157, p. 158, 2010. [5] P. Gasti, G. Tsudik, E. Uzun, and L. Zhang, “Dos and ddos in named data networking,” in 2013 22nd International Conference on Computer Communication and Networks (ICCCN). IEEE, 2013, pp. 1–7. [6] A. Afanasyev, P. Mahadevan, I. Moiseenko, E. Uzun, and L. Zhang, “Interest flooding attack and countermeasures in named data networking,” in 2013 IFIP Networking Conference. IEEE, 2013, pp. 1–9. [7] H. Dai, Y. Wang, J. Fan, and B. Liu, “Mitigate ddos attacks in ndn by interest traceback,” in 2013 IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS). IEEE, 2013, pp. 381–386. [8] A. Compagno, M. Conti, P. Gasti, and G. Tsudik, “Poseidon: Mitigating interest flooding ddos attacks in named data networking,” in 38th annual IEEE conference on local computer networks. IEEE, 2013, pp. 630– 638. [9] T. N. Nguyen, R. Cogranne, G. Doyen, and F. Retraint, “Detection of interest flooding attacks in named data networking using hypothesis testing,” in 2015 IEEE International Workshop on Information Forensics and Security (WIFS). IEEE, 2015, pp. 1–6. [10] H. Salah, J. Wulfheide, and T. Strufe, “Coordination supports security: A new defence mechanism against interest flooding in ndn,” in 2015 IEEE 40th Conference on Local Computer Networks (LCN). IEEE, 2015, pp. 73–81. [11] A. Karami and M. Guerrero-Zapata, “A hybrid multiobjective rbfpso method for mitigating dos attacks in named data networking,” Neurocomputing, vol. 151, pp. 1262–1282, 2015. [12] V. Jacobson, D. K. Smetters, J. D. Thornton, M. F. Plass, N. H. Briggs, and R. L. Braynard, “Networking named content,” in Proceedings of the 5th international conference on Emerging networking experiments and technologies. ACM, 2009, pp. 1–12. [13] S. Mastorakis, A. Afanasyev, and L. Zhang, “On the evolution of ndnSIM: an open-source simulator for NDN experimentation,” ACM Computer Communication Review, Jul. 2017. Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on December 08,2020 at 03:30:32 UTC from IEEE Xplore. Restrictions apply.