IoT data management for caching performance improvement in NDN

doi:10.21203/rs.3.rs-3305411/v1

Download PDF

Research Article

IoT data management for caching performance improvement in NDN

https://doi.org/10.21203/rs.3.rs-3305411/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Connected devices in IoT continuously generate monitoring and measurement data to be delivered to application servers or end-users. Transmitting IoT data through networks would lead to congestion and long delay. NDN is an emerging network paradigm based on name-identified data known to be an appropriate architecture in supporting IoT networks. In-network caching is one of the main advantages of NDN, a major issue discussed in many studies. One of the significant challenges for some IoT data is the transiency nature, making the data caching mechanism different. In this study, the cache placement and data placement are considered together. Data placement is decided upon based on the data lifetime and node position. Data lifetime is an essential property that must be involved in caching methods to prevent invalid data delivery; consequently, the data are classified based on the data lifetime, and specific nodes are selected for caching according to defined classes and nodes position in topology. The objective is to reduce redundancy caused by data caching on all nodes of the data delivery path. By considering both the cache and data placements for transient data, a more comprehensive view is grasped in improving the caching performance. This issue, which has not been addressed in the available studies run on IoT data caching, can lead to the appropriate use of available storage and also reduce redundancy. The obtained results indicate that this method can improve the transient data caching efficiency.

IoT Data

NDN

caching mechanism

lifetime

In the current networks design, often all requests for specific content are directed towards the original data producer, which increases network traffic, especially on the content producer side, content retrieval delay, and bandwidth consumption. The increasing demand for efficient and scalable content distribution has led to introducing of new Information-Centric Network (ICN) architectures[1]. To the researchers, this architecture is an alternative to the existing host-centric networks. In content-centric architecture, The communication model changes from the communication between the two end-nodes into the data-centric model[2, 3].

Unique content naming, addressing, and routing based on the name, decoupling content from its location, applying distributed in-network caching, data packet-based security, and supporting mobility are the obvious advantages of this architecture[4, 5].

Given the challenges in implementing IP-based Internet of Things (IoT), which is often due to the high level of heterogeneity between billions of different devices, the ICN architecture can be applied in implementing it [6, 7]. A variety of IoT devices, including sensors and smartphones, can sense data or exchange them over each other through the internet[8]. In ICN architecture, due to the abstraction of content and its location, heterogeneous devices can connect without middleware[9]. Because ICN supports the main IoT requirements like scalability, energy efficiency, security, big data, and mobility, it is an ideal option in implementing IoT[4, 10, 11].

Different research projects are designed based on the ICN approach. The objective of all ICN architectures is to distribute content and adapt better to disruptions and disconnections in communication[12, 13]. Named Data Networking (NDN) is one of the most common active projects [14], where, the features are most consistent with IoT requirements, thus, the best choice for IoT networks.

The in-network caching technique is mature, and is applied in preventing unnecessary end-to-end communication[15]. The effect of this technique in delivering content to users without requiring direct contact with the original data producer for each request is evident [16]. In-network caching, which is one of the main functionalities of the NDN architecture, is implemented through the CS of nodes[17], Fig. 1. This technique may lead to advantages like cost-effective data retrieval, reduction in latency, heavy load handling, and increase content availability[18, 19].

In default NDN caching, a retrieved content is placed on all the nodes encountered on the path[16]. This placement leads to redundancy for the same content caching. Redundant content caching and limited memory cause inefficiency in the caching system.

The caching studies are of three categories: cache placement, content placement, and replacement policies. Cache placement is related to the determination of where the data will be cached [20]. Content placement is associated with determining which content to cache at the content store[21]. The replacement policy relates to which cached content should be deleted from the content store to accommodate incoming contents[22].

The NDN default caching mechanism achieves a similar caching gain compared to random caching[23]. Researchers in [24] estimated that higher performance is achieved by caching in a set of intermediate routers, but not all nodes of the content delivery path. Accordingly, studies are run to determine a subset of intermediate nodes on the content delivery path, to achieve higher efficiency[23–25]. Many methods are proposed based on the position of nodes and the defined metrics like centrality, degree, betweenness, etc. [24, 26, 27] to overcome this drawback. These methods, following the cache placement strategies category, are based on the properties of nodes and do not consider the properties of the content to be cached.

Due to the inherent characteristics of IoT data and devices, traditional cache strategies for multimedia data do not apply to these systems, which present a new challenge. The large amount of data, on the one hand, and the limited memory space on ICN routers, on the other hand, is another reason why traditional caching strategies are incompatible with IoT systems [28]. A key feature in most IoT data is their being transient[29], that is, in monitoring applications like urban environment monitoring [28, 30], and vehicular traffic monitoring [10, 31], the data are updated on a regular basis. These data are usually characterized by a lifetime within seconds to days range. As long as the defined lifetime by the original data producer is not expired, the data is valid (fresh), otherwise, stale[32]. The studies with a focus on caching transient IoT data are few[29], and the proposed methods, usually with the objective on energy efficiency, due to the limited resources of IoT devices, have not focused on selecting the appropriate set of nodes to reduce redundancy.

This newly proposed method consists of the two primary, cache nodes selective, and content selective sections for IoT data caching. The on-path caching approach is adopted for content caching because, unlike off-path caching, it does not require additional metadata to inform the consumer about the node that is associated with the specified content. In this method, first, the contents are classified based on their lifetimes. And next, they are cached based on their specific node class. The nodes with the highest outgoing interface count or the edge nodes are selected for data caching. By considering both data and cache placement, we determine the suitable caching location for each data class separately. In addition, we remove data that has short lifetime and is not suitable for caching from the caching mechanism. The contributions of this article are:

We propose a novel caching method for transient IoT data that improves the caching performance. One of the significant features for some IoT data is the transiency nature, which we have taken into account in the proposed method.
We involve both node position and IoT data property in this method’s design.
We categorize data through their transient property. Data lifetime is an essential property of multiple IoT data that significantly affects caching performance.

This article is structured as follows: The background and related studies are presented in Section. 2; the method is proposed in Section. 3; the evaluation is made in Section. 4 and the article is concluded in Section. 5.

For different network users, content delivery time is significant. If all requests for the same content are only responded to by the original data producer, this will result in bandwidth consumption, increased delay, and network load. The in-network and transparent caching are applied in NDN architecture to increase content availability and decrease content delivery time through content distribution. Caching is an advantageous and old technique used in traditional networks, but due to the lack of unique content naming, it does not have as many advantages as content-centric networks[18]. In content-centric architecture, with unique and location-independent naming, the content is named the same throughout the network, something applicable in the caching mechanism[13]. In this architecture, the communication is initiated by a user who sends an interest packet and pulls back the data packet by the node, which can be the original producer or a cacher node [33]. Each NDN node has a limited capacity (Content Store) to store the contents that pass through it temporarily[34].

Researchers in [23] revealed that selecting only some of the nodes on the delivery path is more advantageous than always caching; consequently, the following methods are presented accordingly:

In Leave Copy Down (LCD) proposed in [35], only the node at one level down in the reverse path towards the consumer stores a copy. In probabilistic cache (ProbCache) proposed in [36], caching close to consumers is privileged. The caching mechanism is performed with a changing probability inversely proportional to the distance between the consumer and the producer.

The Betweenness Centrality (Btw) strategy proposed by [23] is based on a pre-calculated parameter for each node. This parameter measures the shortest paths’ count between all pairs of nodes in the network going through this node. When the request is sent along a path, first, the highest betweenness centrality of the traversed nodes is stored, and next, it is appended to the data packet. When the response is sent to the consumer, a copy of the content is only cached at the nodes having the highest value. The Edge-caching strategy proposed in [37] is for specific topologies, which caches contents at the topology leaves.

The rationale behind such a selective caching strategy is the fact that some nodes have a higher probability of getting a cache hit than others, and caching the content at better nodes strategically, can decrease the cache eviction rate and increase the overall cache hit ratio.

The other available strategies generally make decisions about caching a particular content based on certain criteria like the content popularity or lifetime.

Researchers in [38] proposed the most popular content strategy by applying a fixed threshold and calculating the number of requests for specific content at any node. When the content is more popular than a fixed threshold, it is cached.

According to [39], the D-FGPC caching strategy is similar to this method, where a variable threshold is applied. For this purpose, the threshold value is calculated and adjusted based on the number of received requests and available storage space[39]. Researchers in [40] made the decision to cache the content based on the node energy level, the space occupied therein, and the content lifetime.

To maximize the energy-efficiency in ICN-based IoT networks, authors of Ref.[41] proposed an Energy-aware Caching Placement (EaCP) scheme. This method tends to tradeoff between content transmission energy and content caching energy by selecting the optimal placement. Its obtained results showed the effectiveness of this strategy in terms of energy saving.

Nour et al proposed a distributed cache placement strategy for large scale information centric network by pushing the popular content at the edge of the network and keep the less popular content at the core of the network[42].

Amado et al presented a caching strategy based on freshness called FDC [43]. They have shown that freshness is an effective parameter in the efficiency of the caching method and can be effective in improving the hit rate and retrieval delay. Considering that data with a short lifetime is regularly removed from the content store of nodes, it is more suitable to store data with a long lifetime. The FDC method defines the probability of data caching using data generation time and data lifetime. These two parameters are added by the original producer in the data packet. When a node receives a data packet, it is cached in the node's content store with probability P(i). If the residual freshness of data packet is greater than the defined threshold, its caching probability is considered one and this data is cached. Otherwise, the probability of data is equal to the ratio of the residual freshness of data to the threshold. The evaluation of the FDC has been done considering a vehicular scenario and compared with the CE2 and the random caching. The comparison results show the superiority of FDC in improving the hit rate and delay. Due to the fact that FDC cache data with a longer residual freshness, the improvement of the hit rate is predictable.

Recently, fog/edge computing with features (e.g., geographically distributed, delay-sensitive, and context-aware) has been considered to deal with the increasing IoT data in many scenarios. Fog computing is used to enhance the quality and security of IoT services in many scenarios. A fog caching scheme enabled by ICN for IoT networks is proposed in Ref.[44]. This scheme may reduce the content delivery time by identifying fog caching nodes near the delivery path. Gupta et al[45] provided better QoE to users by proposing edge caching scheme. In this work, content cached on edge cloud based on its local popularity and predicted future requests. A data prioritization-based approach that prevents ICN nodes from filling their Content Store (CS) with inappropriate and inferior data is proposed in Ref.[46]. In this work, fog computing is used as a middle layer between IoT devices and ICN networks to resolve the limitations of ICN routers and improve cache performance.

The studies where the features of NDN-based IoT systems are of concern are few. According to [14], due to the characteristics of IoT systems, these caching strategies are ineffective for NDN-based IoT systems. Most of the presented methods are for large and non-transient multimedia files that are frequently requested by users. Recently the possibility of caching transient IoT data and caching these data is discussed and supported by ICNRG subject to the Internet Research Task Force (IRTF)[25].

In this paper, we focus on caching transient IoT data with respect to reduce redundancy. Since a lot of data is often transmitted in IoT networks and the cache capacity of routers is often limited, proper decision making is necessary to place the appropriate data in suitable nodes for in-network caching. So, we present a model for data caching in ICN-based IoT using data classification to manage the data and improve the cache efficacy. We categorize data based on data lifetime, because most of the IoT applications required fresh data with respect to time constraints and most of the IoT data are transient and life-limited in nature that needs to be replaced. Paying attention to the transient nature of most IoT data can improve caching performance. Given this property and the location of the nodes in the network, we present a data caching method for ICN-based IoT. According to our studies, the existing works have mainly focused on the location of nodes and other data properties such as popularity.

The attempt is made to propose a new caching scheme with the beyond-mentioned features in mind, first, this caching scheme will function at locations with a high probability of getting cache hits, and next, it will cache the fresher content items based on its lifetime. The main modules of this proposed method are shown in Fig. 2.

3.1 Data classification

IoT data have distinctive characteristics that make traditional caching schemes obsolete. A significant volume of heterogeneous, streaming, and geographically dispersed real-time data is being generated by numerous devices, which forward observations about specific monitored phenomena or reporting the occurrence of certain or abnormal events periodically [47].

The data producer adds a new lifetime field to the data packet, indicating the data validity time after it is generated[16]. Data updating can be periodic (at regular intervals) or non-periodic. Moreover, a timestamp field is added to the data packet to indicate when this data item is generated [32]. According to these two fields, the real-time where this data would expire is calculated as follows:

$${t}_{exp}=T+ {t}_{s}$$

Where,${t}_{s}$, T, and${t}_{exp}$ are the data generation time by the original producer, lifetime, and actual expiration time, respectively. Depending on the data receipt time $\left({t}_{arr}\right)$ and${t}_{exp}$, it is determined whether they are fresh or stale. If${t}_{arr}$ for specific data is less than${ t}_{exp}$, this data is fresh, otherwise stale. The terms freshness second, and lifetime mean the same and are often applied interchangeably [48].

The data are categorized through their transient characteristic. Because data lifetime is a fundamental parameter for IoT data caching, within a few seconds to several hours range, the available data is categorized into three classes, Table.1.

Class I contains very transient data, like security-related information or information about an acute patient's condition, generated with a very short lifetime by the producer. In general, this type of data is not very popular and is only requested by very few users in certain applications.

Class II is defined as the data with a lifetime less than that of the T1. Class II refers to data with a normal lifetime, like traffic information for a given environment and road congestion. This class contains data that its specified lifetime is within T1 to T2 range.

Class III refers to data with a long lifetime, which can be valid beyond T2 to several hours, as some information about road routes, the weather conditions of an area, or commercial and advertising services.

Classes II and III are often applied by more consumers than Class I, (i.e., more popular). Therefore, by classifying the data based on their lifetime, an appropriate and separate caching mechanism can be applied for each class, which would lead to caching mechanisms’ improvement.

Table 1

Data Classification
Data classification	Lifetime range	Appropriate data
Class I : Short living	T ≤ T1	Frequently updated data
Class II : Normal living	T1 < T ≤ T2	Normal data
Class III: Long living	T > T2	Static data

3.2Cache selective mechanism

The major advantage of applying the NDN in implementing IoT is its storage capacity in decreasing latency and increasing data availability when data is cached on intermediate routers. Data caching in all nodes of the path leads to redundancy. It has the same and even less efficiency compared to random caching in one of the path nodes[23]. So, we select the most important nodes for caching based on the location of nodes. The data is classified based on its lifetime for caching each class separately. As shown in Fig. 3, data caching proposed based on the nodes position and the specified classes.

Class I: Short living data

The storage space of NDN routers is limited compared to the significant volume of IoT data. A recent report from IoT Analytics predicts that the connected devices would grow to about 34.1 billion by 2025 [49]. These devices produce great volumes of data coming from the array of embedded sensors and controllers. According to[50], the maximum available memory for modern routers can be 10 Gbyte, and according to [51], this is possible at a high cost. The volume of the data transmitted over the nowadays network is much greater than 10G[52].

Short living data is updated frequently and is often requested only in specific applications by limited consumers. Due to the limitations in memory, the data of this class is not cached. So, the available cache is used for caching the data with a longer lifetime to improve caching performance.

Class II: Normal living data

In these data, with a relatively short lifetime, data is cached in the edge nodes. The findings in [20, 37] indicate that edge nodes constitute an efficient option for data caching. In this process, due to the proximity to consumers and the limited data lifetime, access to data before expiration by reducing delay becomes possible. Due to the limited lifetime of this data class, edge nodes are selected as an effective position to cache them.

Class III: Long living data

Long-living data are the most appropriate for caching. Due to the centrality metrics, which are based on the position of the nodes, caching nodes for this data class is selected. These metrics are also used in social networks to determine the important nodes, the nodes in the network that are highly connected in their topological sense, and include betweenness, degree, closeness, etc.[53].

In algorithms where the degree metric is applied, data is often cached at nodes on the return path, with a high count of direct connections. High-degree nodes are connected to a great count of nodes in their immediate region. The closeness metric of a specific node determines the average length of the shortest paths between this node and all others in the network. Data caching is made on the nodes on the path with high closeness. For each node, the betweenness is determined based on the shortest path count between all pairs of nodes in the network crossing that node. In betweenness-based algorithms, the node with the highest value in the data delivery path caches the data[4, 53, 54].

In this study, the betweenness metric is applied for long-living data, where more centralized nodes cache data. Some nodes are in contact with a highly centered node. By storing data in these nodes, more cache hit occurs for the consumers' requests that pass through these nodes, while, many requests are satisfied in a shorter delay. This metric is defined through Eq. (2), [24]:

$$Bet\left(x\right)=\sum _{i\ne j,i,j\in V-\left\{x\right\}}\frac{n(i,j,x)}{n(i,j)}$$

Where $n(i,j)$ is the total content delivery path count between two nodes i, j, and $n\left(i,j,x\right)$ is the count of the paths that pass through node x. Node i requests the content, and node j is either the original content producer or the node caching the content.

In this proposed method, short-living data is not cached in NDN nodes, thus, a reduction in the total numbers of data packets that can be cached. The effect of reducing the data count on the network performance metric related to caching, namely the data latency and the hit rate, is evaluated with the assumption that T1 = 2s and T2 = 6s. Class I is defined as the data with a lifetime shorter than two seconds. Class II contains data that its specified lifetime is within two seconds to six seconds range. Class III refers to long lifetime data, with a lifetime within six seconds to more seconds up to several hours.

The open-source NDNsim package, a data-centric network simulator provided for NDN architecture, is applied for simulation. This simulator is implemented on the NS-3 platform, a free network simulation tool [55].

The simulated scenario is run to materialize the objective of this study. This simulation consists of 20 consumers arranged on the tree topology, where each consumer is directly connected to a leaf node. The number of data is considered as a variable. In this simulation, any consumer can request any data. The consumers submit requests at the rate of 10 requests per second. The data consumer and producer are considered to be no-caching to prevent requests from being satisfied by their internal cache. The cache size was defined as the same for all intermediate routers. The simulation parameters of this scenario are briefly listed in Table 2. For evaluation, two metric are considered in two different scenario: cache hit ratio and content retrieval delay.

Table 2

Simulation Parameters
Number of consumers	20
Content size	10 packets
Cache size	30 packets
Simulation time	100 s

The effect of reducing the number of data on latency and cache hit ratio with three different replacement policies are shown in Fig. 4 and Fig. 5, respectively. LRU, random, and LFU replacement policies had been applied for this evaluation. Based on the findings, it is clear that by removing the first-class data from the caching mechanism, the hit ratio increases, and delay time decreases. All three applied replacement policies have similar behavior, leading to reduced hit rates and increased latency due to increased content, while, better results are obtained by adopting the LRU policy.

On the one hand, vast amounts of data are generated by IoT devices in most real IoT systems. On the other hand, all the generated data are not of the same importance in terms of the caching. Therefore, by identifying low important data and removing those from the caching mechanism, and ultimately reducing the number of cached data, significant improvements in caching performance can be achieved.

In the proposed method, we incorporate an essential data characteristic, data lifetime, into the caching mechanism. In Fig. 6, graphs for different values of this parameter are presented to observe the changing behavior of the cache hit ratio. The results of varying lifetime show that the longer the data lifetime seconds, the more likely the intermediate routers will respond to requests. In other words, as the data lifetime increases, the cache hit ratio increases.

To assess the effect of data freshness on the caching methods, another scenario is run. In this scenario, one producer and one consumer are implemented in the grid topology. The cache size is limited to 20 packets. The LRU is applied as the cache replacement policy.

In this scenario, the consumer begins requesting each content at specific times. The hit ratio values for data with a lifetime less than two seconds are almost zero, indicating that, as expected, their caching is not necessary in the simulation. This allows the removal of these data from the caching mechanism, and provides the possibility of caching more appropriate data, the data with a higher lifetime.

Finally, we consider a content catalog of 1000 transient contents. In this scenario, all content sizes were assumed to be the same. The cache size was defined as the same for all intermediate routers. The simulated scenario was designed to facilitate the achievement of our research objectives. It comprises consumers arranged on binary tree topology, and each consumer was directly connected to a leaf node. Also, any consumer could request any data. Consumers request contents according to the Zipf’s law, with skewness parameter set to 1. Zipf distribution is commonly used to model content popularity in the NDN networks. Based on this distribution, the access probability of the k^th content is defined as:

$$F\left(k;\alpha ,M\right)=\frac{1/{k}^{\alpha }}{{\sum }_{j=1}^{M}\frac{1}{{j}^{\alpha }}}$$

Where a is denoted as skewness parameter, M is content number.

We also considered data consumers and producers to be no-caching to prevent requests from being satisfied by their internal cache. A variable number of consumers, ranging from 20 to 100, is assumed in this scenario.

The results of average delay and hit ratio in data caching when varying the number of consumers are shown in Figs. 7 and Fig. 8, respectively. The proposed method is compared against two strategies, namely CE² and Betweenness centrality (Btw)[23][14]. In all these methods, LRU is used as a replacement policy. We assumed that the data producer generates data with different freshness. The data freshness is a random value in the range [0.5-8] seconds.

The Betweenness centrality (Btw) caching method cached data on only one of the intermediate nodes in the path between producer and consumer. Details of this method are given in Section 2. As expected, in all the considered cases, the result represented that the metric exhibits lower values as the number of consumers increases. This is because is a higher chance to request more distinct contents.

A caching strategy based on transient IoT data classification is proposed. Some of the IoT data are transient, and this feature makes data caching vary. Also, the freshness is a critical challenge particularly in IoT environment. In this proposed method, the three classes of data based on the data lifetime are defined. The data placement based on the position of the nodes and the specified classes are. By selecting the set of nodes, the caching nodes’ number is reduced instead of data caching on all nodes of the path. This reduction prevents redundancy caused by data storing in all nodes of the data transfer path. The data of class II and III are cached only at the edge nodes and node with the highest Bet(x) in the path, respectively. This proposed strategy is evaluated by running two scenarios. Performance assessments obtained with ndnsim have demonstrated the superiority of the proposed method in terms of cache hit ratio and content retrieval latency against two schemes available in the literature. Due to the limited caching space in routers, this proposed method contributes to improving the caching performance. Data distribution, data frequency of each class, and data lifetime intervals applied to classify data constitute the effective parameters in evaluating this method. These parameters require further assessments.

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

I.U. Din, H. Asmat, M. Guizani, A review of information centric network-based internet of things: communication architectures, design issues, and research opportunities, Multimed. Tools Appl. 78 (2019) 30241–30256. https://doi.org/10.1007/s11042-018-6943-z.
D. Mars, S. Mettali Gammar, A. Lahmadi, L. Azouz Saidane, Using Information Centric Networking in Internet of Things: A Survey, Wirel. Pers. Commun. 105 (2019) 87–103. https://doi.org/10.1007/s11277-018-6104-8.
A. Lindgren, F. Ben Abdesslem, B. Ahlgren, O. Schelén, A.M. Malik, Design choices for the IoT in Information-Centric Networks, 2016 13th IEEE Annu. Consum. Commun. Netw. Conf. CCNC 2016. (2016) 882–888. https://doi.org/10.1109/CCNC.2016.7444905.
F. Khandaker, S. Oteafy, H.S. Hassanein, H. Farahat, A functional taxonomy of caching schemes: Towards guided designs in information-centric networks, Comput. Networks. 165 (2019). https://doi.org/10.1016/j.comnet.2019.106937.
A. V. Vasilakos, Z. Li, G. Simon, W. You, Information centric network: Research challenges and opportunities, J. Netw. Comput. Appl. 52 (2015) 1–10. https://doi.org/10.1016/j.jnca.2015.02.001.
S. Arshad, M.A. Azam, M.H. Rehmani, J. Loo, Recent advances in information-centric networking-based internet of things (ICN-IoT), IEEE Internet Things J. 6 (2019) 2128–2158. https://doi.org/10.1109/JIOT.2018.2873343.
J. Suarez, J. Quevedo, I. Vidal, D. Corujo, J. Garcia-Reinoso, R.L. Aguiar, A secure IoT management architecture based on Information-Centric Networking, J. Netw. Comput. Appl. 63 (2016) 190–204. https://doi.org/10.1016/j.jnca.2016.01.016.
M. binti Mohamad Noor, W.H. Hassan, Current research on Internet of Things (IoT) security: A survey, Comput. Networks. 148 (2019) 283–294. https://doi.org/10.1016/j.comnet.2018.11.025.
B. Nour, K. Sharif, F. Li, H. Moungla, A.E. Kamal, H. Afifi, NCP: A near ICN Cache Placement Scheme for IoT-Based Traffic Class, in: 2018 IEEE Glob. Commun. Conf. GLOBECOM 2018 - Proc., 2018. https://doi.org/10.1109/GLOCOM.2018.8647629.
E. Baccelli, C. Mehlis, O. Hahm, T.C. Schmidt, M. Wählisch, Information centric networking in the IoT: Experiments with NDN in the wild, ICN 2014 - Proc. 1st Int. Conf. Information-Centric Netw. (2014) 77–86. https://doi.org/10.1145/2660129.2660144.
R. Ullah, S.H. Ahmed, B.S. Kim, Information-centric networking with edge computing for IoT: Research challenges and future directions, IEEE Access. 6 (2018) 73465–73488. https://doi.org/10.1109/ACCESS.2018.2884536.
Z. Ali, M.A. Shah, A. Almogren, I. Ud Din, C. Maple, H.A. Khattak, Named data networking for efficient IoT-based disaster management in a smart campus, Sustain. 12 (2020) 1–21. https://doi.org/10.3390/SU12083088.
B. Ahlgren, C. Dannewitz, C. Imbrenda, D. Kutscher, B. Ohlman, Networking ( Draft )A Survey of Information-Centric, Dagstuhl ICN Work. (2011) 1–26. https://doi.org/10.1109/MCOM.2012.6231276.
M. Meddeb, A. Dhraief, A. Belghith, T. Monteil, K. Drira, S. Alahmadi, Cache freshness in named data networking for the internet of things, Comput. J. 61 (2018) 1496–1511. https://doi.org/10.1093/comjnl/bxy005.
Y. Kim, I. Yeom, Performance analysis of in-network caching for content-centric networking, Comput. Networks. 57 (2013) 2465–2482. https://doi.org/10.1016/j.comnet.2012.11.026.
S. Vural, N. Wang, P. Navaratnam, R. Tafazolli, Caching Transient Data in Internet Content Routers, IEEE/ACM Trans. Netw. 25 (2017) 1048–1061. https://doi.org/10.1109/TNET.2016.2616359.
S. Kumar, R. Tiwari, S. Kozlov, J.J.P.C. Rodrigues, Minimizing delay in content-centric networks using heuristics-based in-network caching, Cluster Comput. (2022). https://doi.org/10.1007/s10586-021-03405-1.
G. Zhang, Y. Li, T. Lin, Caching in information centric networking: A survey, Comput. Networks. 57 (2013) 3128–3141. https://doi.org/10.1016/j.comnet.2013.07.007.
I. Abdullahi, S. Arif, S. Hassan, Survey on caching approaches in Information Centric Networking, J. Netw. Comput. Appl. 56 (2015) 48–59. https://doi.org/10.1016/j.jnca.2015.06.011.
M. Meddeb, A. Dhraief, A. Belghith, T. Monteil, K. Drira, How to cache in ICN-based IoT environments?, Proc. IEEE/ACS Int. Conf. Comput. Syst. Appl. AICCSA. 2017-Octob (2018) 1117–1124. https://doi.org/10.1109/AICCSA.2017.37.
J. Pfender, A. Valera, W.K.G. Seah, Content delivery latency of caching strategies for information-centric IoT, ArXiv. (2019) 1–10.
A. Kalla, S.K. Sharma, A constructive review of in-network caching: A core functionality of ICN, in: Proceeding - IEEE Int. Conf. Comput. Commun. Autom. ICCCA 2016, 2017: pp. 567–574. https://doi.org/10.1109/CCAA.2016.7813785.
W.K. Chai, D. He, I. Psaras, G. Pavlou, Cache “less for more” in information-centric networks, in: Int. Conf. Res. Netw., Springer, 2012: pp. 27–40.
K.N. Lal, A. Kumar, A Centrality-measures based Caching Scheme for Content-centric Networking (CCN), Multimed. Tools Appl. 77 (2018) 17625–17642. https://doi.org/10.1007/s11042-017-5183-y.
S. Vural, P. Navaratnam, N. Wang, C. Wang, L. Dong, R. Tafazolli, In-network caching of Internet-of-Things data, in: 2014 IEEE Int. Conf. Commun., IEEE, 2014: pp. 3185–3190.
Y. Cui, M. Zhao, M. Wu, A centralized control caching strategy based on popularity and betweenness centrality in CCN, Proc. Int. Symp. Wirel. Commun. Syst. 2016-Octob (2016) 286–291. https://doi.org/10.1109/ISWCS.2016.7600916.
J.A. Khan, C. Westphal, Y. Ghamri-Doudane, A Popularity-aware Centrality Metric for Content Placement in Information Centric Networks, 2018 Int. Conf. Comput. Netw. Commun. ICNC 2018. (2018) 554–560. https://doi.org/10.1109/ICCNC.2018.8390396.
S.R. Jayarajanneditors, EAI/Springer Innovations in Communication and Computing Integration offWSN and IoT forrSmart Cities, 2020. https://doi.org/10.1007/978-3-030-38516-3.
M.A. Naeem, R. Ullah, Y. Meng, R. Ali, B.A. Lodhi, Caching Content on the Network Layer: A Performance Analysis of Caching Schemes in ICN-Based Internet of Things, IEEE Internet Things J. 9 (2022) 6477–6495. https://doi.org/10.1109/JIOT.2021.3110977.
O. Zedadra, A. Guerrieri, N. Jouandeau, H. Seridi, G. Fortino, G. Spezzano, The Internet of Things for Smart Urban Ecosystems, Springer International Publishing, 2019. https://doi.org/10.1007/978-3-319-96550-5.
Z. Zhang, C.H. Lung, I. Lambadaris, M. St-Hilaire, IoT data lifetime-based cooperative caching scheme for ICN-IoT networks, IEEE Int. Conf. Commun. 2018-May (2018). https://doi.org/10.1109/ICC.2018.8422100.
J. Quevedo, D. Corujo, R. Aguiar, Consumer driven information freshness approach for content centric networking, Proc. - IEEE INFOCOM. (2014) 482–487. https://doi.org/10.1109/INFCOMW.2014.6849279.
I.U. Din, S. Hassan, M.K. Khan, M. Guizani, O. Ghazali, A. Habbal, Caching in Information-Centric Networking: Strategies, Challenges, and Future Research Directions, IEEE Commun. Surv. Tutorials. 20 (2018) 1443–1474. https://doi.org/10.1109/COMST.2017.2787609.
S.N. Seyyed Hashemi, A. Bohlooli, Analytical characterization of cache replacement policy impact on content delivery time in information-centric networks, Int. J. Commun. Syst. 32 (2019) 1–13. https://doi.org/10.1002/dac.4154.
N. Laoutaris, H. Che, I. Stavrakakis, The LCD interconnection of LRU caches and its analysis, Perform. Eval. 63 (2006) 609–634. https://doi.org/10.1016/j.peva.2005.05.003.
I. Psaras, W.K. Chai, G. Pavlou, Probabilistic in-network caching for information-centric networks, in: Proc. Second Ed. ICN Work. Information-Centric Netw., 2012: pp. 55–60.
S.K. Fayazbakhsh, Y. Lin, A. Tootoonchian, A. Ghodsi, T. Koponen, B. Maggs, K.C. Ng, V. Sekar, S. Shenker, Less pain, most of the gain, ACM SIGCOMM Comput. Commun. Rev. 43 (2013) 147–158. https://doi.org/10.1145/2534169.2486023.
C. Bernardini, T. Silverston, O. Festor, MPC: Popularity-based caching strategy for content centric networks, IEEE Int. Conf. Commun. (2013) 3619–3623. https://doi.org/10.1109/ICC.2013.6655114.
D.O. Mau, M. Chen, T. Taleb, X. Wang, V.C.M. Leung, FGPC: Fine-Grained Popularity-based Caching design for Content Centric Networking, in: MSWiM 2014 - Proc. 17th ACM Int. Conf. Model. Anal. Simul. Wirel. Mob. Syst., 2014. https://doi.org/10.1145/2641798.2641837.
M.A. Hail, M. Amadeo, A. Molinaro, S. Fischer, Caching in named data networking for the wireless internet of things, in: 2015 Int. Conf. Recent Adv. Internet Things, IEEE, 2015: pp. 1–6.
O. Serhane, K. Yahyaoui, B. Nour, H.M.E.C. Place-, O. Serhane, K. Yahyaoui, B. Nour, H. Moungla, Energy-aware Cache Placement Scheme for IoT-based ICN Networks, (2021).
B. Nour, H. Khelifi, H. Moungla, R. Hussain, N. Guizani, A Distributed Cache Placement Scheme for Large-Scale Information-Centric Networking, IEEE Netw. 34 (2020) 126–132. https://doi.org/10.1109/MNET.011.2000081.
M. Amadeo, C. Campolo, G. Ruggeri, G. Lia, A. Molinaro, Caching Transient Contents in Vehicular Named Data Networking: A Performance Analysis, Sensors. 20 (2020) 1985. https://doi.org/10.3390/s20071985.
Y. Hua, L. Guan, K.G. Kyriakopoulos, A Fog caching scheme enabled by ICN for IoT environments, Futur. Gener. Comput. Syst. 111 (2020) 82–95. https://doi.org/10.1016/j.future.2020.04.040.
D. Gupta, S. Rani, S.H. Ahmed, S. Verma, M.F. Ijaz, J. Shafi, Edge caching based on collaborative filtering for heterogeneous icn-iot applications, Sensors. 21 (2021) 1–17. https://doi.org/10.3390/s21165491.
M.S. Zahedinia, M.R. Khayyambashi, A. Bohlooli, Fog-based caching mechanism for IoT data in information centric network using prioritization, Comput. Networks. 213 (2022) 109082. https://doi.org/10.1016/j.comnet.2022.109082.
M. Abu-Elkheir, M. Hayajneh, N.A. Ali, Data management for the Internet of Things: Design primitives and solution, Sensors (Switzerland). 13 (2013) 15582–15612. https://doi.org/10.3390/s131115582.
M. Amadeo, G. Ruggeri, C. Campolo, A. Molinaro, G. Mangiullo, Caching popular and fresh IoT contents at the edge via named data networking, in: IEEE INFOCOM 2020 - IEEE Conf. Comput. Commun. Work. INFOCOM WKSHPS 2020, 2020: pp. 610–615. https://doi.org/10.1109/INFOCOMWKSHPS50562.2020.9162741.
Statista.com, • IoT: number of connected devices worldwide 2012-2025 | Statista, Web. (2018).
S. Arianfar, P. Nikander, J. Ott, On content-centric router design and implications, in: Proc. Re-Architecting Internet Work. Held Conjunction with Conex. 2010, 2010. https://doi.org/10.1145/1921233.1921240.
M. Wang, J. Wu, G. Li, J. Li, Q. Li, S. Wang, Toward mobility support for information-centric IoV in smart city using fog computing, in: 2017 5th IEEE Int. Conf. Smart Energy Grid Eng. SEGE 2017, 2017: pp. 357–361. https://doi.org/10.1109/SEGE.2017.8052825.
M. Wang, J. Wu, G. Li, J. Li, Q. Li, Fog computing based content-aware taxonomy for caching optimization in information-centric networks, 2017 IEEE Conf. Comput. Commun. Work. INFOCOM WKSHPS 2017. (2017) 474–475. https://doi.org/10.1109/INFCOMW.2017.8116422.
A. Kalla, S.K. Sharma, A constructive review of in-network caching: A core functionality of ICN, Proceeding - IEEE Int. Conf. Comput. Commun. Autom. ICCCA 2016. (2017) 567–574. https://doi.org/10.1109/CCAA.2016.7813785.
J.A. Khan, C. Westphal, Y. Ghamri-Doudane, A content-based centrality metric for collaborative caching in information-centric fogs, 2017 IFIP Netw. Conf. IFIP Netw. 2017 Work. 2018-Janua (2017) 1–6. https://doi.org/10.23919/IFIPNetworking.2017.8264869.
S. Mastorakis, A. Afanasyev, I. Moiseenko, L. Zhang, ndnSIM 2 . 0 : A new version of the NDN simulator for NS-3, NDN Proj. (2015) 1–8.

No competing interests reported.

Download PDF

Editorial decision: Major revision
14 Oct, 2023
Reviews received at journal
09 Sep, 2023
Reviewers agreed at journal
01 Sep, 2023
Reviewers invited by journal
01 Sep, 2023
Editor assigned by journal
31 Aug, 2023
Submission checks completed at journal
29 Aug, 2023
First submitted to journal
29 Aug, 2023

You are reading this latest preprint version

IoT data management for caching performance improvement in NDN

Status:

Version 1

Abstract

Figures

1. Introduction

2. Related works

3. Proposed method

3.1 Data classification

3.2Cache selective mechanism

4. Evaluation

5. Conclusion

Declarations

Competing interests

References

Additional Declarations

Status:

Version 1