Analytical modelling solution of producer mobility support scheme for named data networking

Received Oct 3, 2018 Revised Apr 2, 2019 Accepted Apr 14, 2019 Named Data Networking (NDN) is a clean-slate future Internet architecture proposed to support content mobility. However, content producer mobility is not supported fundamentally and faces many challenges such as high handoff latency, signaling overhead cost and unnecessary Interest packet losses. Hence, many approaches indirection-based approach, mapping-based approach, locator-based approach and control/data plane-based approach were proposed to address these problems. Mapping-based and control/data plane-based approach deployed servers for name resolution services to provide optimal data path after the handoff, but introduces high handoff latency and signaling overhead cost. Indirection-based and locator-based approach schemes provide normal handoff delay but introduce sub-optimal or triangular routing path. Therefore, there is needs to provide substantial producer mobility support that minimizes the handoff latency, signaling cost and improve data packets delivery via optimal path once a content producer relocates to a new location. This paper proposed a scheme that provides optimal data path using mobility Interest packets and broadcasting strategy. Analytical investigation result shows that our proposed scheme outperforms existing approaches in terms of handoff latency, signaling cost and path optimization.


INTRODUCTION
Named Data Networking (NDN) is an architecture completely re-design for the proposed future Internet that evolved principally from the successes of the current Internet that are IP architecture [1,2]. The architecture was proposed to supplement and replace IP architecture due to the high demand of network content such as mobile data [3], high bandwidth and mobility [4] latency in IoT network [5], mobility management and vehicular networks VANET [6]. To achieve routing and forwarding of packets, NDN nodes uses named-base routing and possesses two different types of packets, namely Interest and Data packets. The node can be represented as client consumer, producer or a router, that maintains three data structure Pending Interest Table (PIT), Forwarding Information Base (FIB) and Content Store (CS), determines when and where to forward Data and Interest packets [7]. PIT records and store any incoming Interest information, FIB maintained the Forwarding Strategy and decide when and where to forward Interest and CS is a temporary cache of data stored based on the NDN caching policy.
Fundamentally, NDN was proposed to support content mobility, thus proposed as a consumerdriven network. When content consumer relocates to a new domain or access point, it has the ability to send pending Interest packets that are not satisfied with relevant data. The caching capability of NDN suppressed The Indirection-based Mobility Approach (IBMA) is a technique derived from the MIPv4 and MIPv6 mobility solution concept to support the mobile producer in NDN architecture [11]. A home agent or home router was provided to maintain the binding information between content prefix content location and redirect the Interest packets using the tunnel to the new location or PoA of the mobile producer [29,30]. Lee et al. also proposed an indirection approach similar to [30] that uses the home domain Content Router (CR h ) and intermediate Content Routers (CR) to handle Producer mobility. The mobile producer always sends a prefix update to CR h about its movement to a new location; also announce the prefix to the new CR [29]. A proposed protocol was design, evaluated and benchmark against purely indirection approach and traditional CCN architecture. The names were used to update the home router and old PoA about the mobility event of content producer [31,32]. For the normal process when the producer does not move, all interest packets were forwarded through the home router using static FIB. Yan et al. proposed a distributed mobility management scheme as a solution for both consumer and producer mobility in NDN architecture [33], and all mobile IP architecture supported by NDN overlay [28]. The distributed mobility management handover scheme supports both consumer and producer mobility which minimized handoff latency and routing update cost or signaling overhead cost. In general, the indirection-based approach provides normal handoff latency and signaling cost. However, the approach has a triangular routing problem.
Rao, Luo et al. [16] and Rao Gao et al. [15] proposed Locator-Based producer mobility scheme by adding a locator to each AR (Access Router) in NDN network as unique topological layer, also extend the functionality of AR with capability of catching, forwarding Interest packets on behalf of the source and added a field in the original NDN packet. FIB entries that contain the only list of interfaces is modified to include a list of mobility status and locator. The scheme was called locator/identifier separation-based approach [19] where the mapping of producer's content prefix that serves as an identifier and its current location take place by home router. Location-aware on-demand multipath protocol based on NDN MANETs was proposed by Asif and Kim [34] to solve the intermittent connectivity, low battery power, data redundancy, packet flooding that affect network performance due to the nature of broadcast nature of wireless mobile ad-hoc network in NDN. In the study of Azgin, Ravindran and Wang an anchor chain on-demand mobility support solution for ICN was proposed to manage producer mobility and handoff processes to avoid data loss [35]. The architecture used decentralized micro-level resolution system and chained distributed anchors that provide forwarding functionalities to help for the efficient packets forwarding to the mobile producer after handoff [35]. In general, LBMA provides normal handoff latency, signaling cost and partial modification of NDN architecture. However, the approach did not consider path optimization after the handoff and the central node (Home repository, Anchor, Access router) utilization can be resulted in path stretching and serve as a single point of failure.
The Control/Data Plane-based Mobility Approach (CDBMA) is a technique based on the control plane and data plane separation to support mobile producer. The control plane used a server that serves as controller [36] or resource handler [37] to control the process of mobility signaling during the handoff. To reduce the handoff latency and minimized the signaling overhead Ren et al. proposed a producer mobility management scheme using Software Define Controller (SDC) to improve the performance of [37] scheme. The basic idea of the proposed scheme is to separate control from data plane as in [36,37], updating routing tables entries and configures the relevant content routers by the SDC [38]. Joao et al. proposed Controllerbased Routing Scheme for NDN to solve the scalability problems caused by content mobility that intensifies a high number of contents in different locations. The scheme splits content identity and content location from name prefix to facilitate content mobility [39]. However, the scheme does not ensure path optimization after the handoff and when producer mobility rate increases; the overhead will increase and degrade the efficiency. In general, CDBMA provides low packets delivery cost, However, the approach is more of software-defined networking architecture, which is not suitable with NDN architecture and the handoff latency and signaling cost was very high due to the consumer, producer querying processes to the resolution handler server and entire FIBs update.

PROPOSED PRODUCER MOBILITY SUPPORT SCHEME
Content producer movement in NDN causes some problems that need to be addressed, such problems are high handoff latency or delay, high signaling cost, routing path sub-optimization and unnecessary Interest packet lost. When producer moved to new PoA, the consumer, on the other hand, will continue sending Interest packets via the old path, towards the earlier location of the producer as normal transmission, without knowledge of the producer movement. The PoA that previously attached to the producer will continuously drop the packets. Even though caching plays an important role in the NDN network, but there are certain consumer's requests that are having many numbers of data chunks. Real-time communication that has less priority in caching, or the unsatisfied consumer request, that will make the

Mobility interest package
In addition, a new packet Mobility Interest (MI) will be introduced to carry binding information for the content identifier, old and new location of mobile producer, to update the immobile anchor router. The MI packet is similar in general purpose with different functions and format to the forwarding hint [12,16], binding update [12,29], mobility management packet [19,26,31], traced and tracing Interest [40] etc. However, the packets are differed based on the different schemes and techniques.

Broadcast strategy
When mobile producer relocates to new PoA there is need to update the network about the new name prefix of producer's new location. Zhu et al. [13] in NDN technical report that highlighted the new perspective on mobility support in NDN suggested that the mapping between content identifier and content locator can be provided by the broadcasting method of using intermediate nodes like DNS servers. Added that, Interests are relatively small and NDN is broadcast friendly making it practically possible to broadcast in a restricted domain to track the new location of the producer. For a large network, broadcast overlay spanning can be used [34]. NDN support Interest Broadcasting as the Interest are small in size making it feasible to be broadcasted in the restricted domain to track the location of mobile producer [13]. Figure 1 shows the architecture and mode of operation of the proposed scheme.

FORMULATION OF ANALYTICAL MODEL 4.1. Network analysis model
To analyze and evaluate the mobility handoff performance of the proposed scheme, we consider a network analysis model as done in [16,19,23,41]. The comparison of the proposed scheme and existing were addressed using network analytical model to validate the concept of broadcasting strategy. Handoff latency and cost were been mathematically formulated for MBMA, LBMA, IBMA and CDPBMA to evaluate the concept and address handoff problems. Mathematical models are easier to represents logical behavior between network elements, also covers the bound of research question by showing the borderline of the network behavior or characteristics. The notation, parameters, and values used are presented in Table 1 and the build network analysis model is shown in Figure 2. In the network analysis model presented, the hop count between the content router and consumer or producer, content routers and server represented by a, b, c and d as shown in Table 1. We assumed that the producer has a constant speed.

Handoff latency analysis
Handoff or handover is a term used interchangeably in mobile networking that refers to the processes where a mobile producer disconnects from current PoA and connects to new neighboring PoA. The period of time that content producer requires from the last Packets (Interest or Data) received via old PoA to the arrival of the first Packets via the new PoA after the handoff is called handoff latency the smaller latency, the better performance of the proposed scheme. (1) and (2) is the wired and wireless link delay between two hops for data, Interest, update or query as used in [19]. Where q is the probability of link failure, B w and B wl are wired and wireless bandwidth, Ld w and Ld wl link delay for wired and wireless and Q d is the queuing delay. Their values, B w =100 Mbps, B wl= 11 Mbps, Ld w =2 ms and Ld wl= 10 ms and Q d= 5 ms The handoff latency of MBMA can be express as (4). When the content producer moves to new AR, new prefix name should be generated, and the new AR sent an updated prefix to the server. The server performs mapping process to update existing name prefix to generate a forwarding hint or binding update. When round trip time (RTT) elapsed, the content consumer query the server requesting the forwarding hint that will be used to trace the new location of the content producer. The total handoff delay can be generated in (3) together with the two consecutive delays for wired and wires link defined in (1) The handoff latency of LBMA can be express as (4). The handoff process takes place when the producer moves to new AR, after generating new prefix name the new AR sent a location update or binding update to the home router or old AR. When the handoff procedure is over, the content consumer will send the Interest to the old AR that will encapsulate and forward encapsulated Interest to the new AR and the producer. The total handoff delay can be generated as follows: The handoff latency of CDBMA is closely related to MBMA and can be express as (5). When the content producer moves to new AR, new prefix name should be generated, and the new AR sent an updated prefix to the control server. The control server performs a mapping process and updates the whole content router (ARs) in the domain. The content consumer can request a control server for the forwarding hint if its AR was not up-to-date. The total handoff latency can be generated as follows: The handoff latency of IBMA can be express as (6), is also like that of LBMA, but the binding update is used to update the FIBs of both old and new AR. When the handoff procedure is over, the content consumer will send the Interest to the old AR that will redirect the Interest to the new location of the producer. The total handoff latency can be generated as follows: The handoff latency of propose scheme (PMSS) can be express as (7). When producer completes the handoff between old AR and new AR, it sends its new prefix information in MI packet to the new access router and the new access router forward it to the known location of the immobile anchor. The immobile anchor broadcast the MI to update the intermediate routers in the restricted domain. Then a consumer can reissues pending Interest toward the new location of the content producer. The expression is as follows:

Handoff signaling and packets delivery cost analysis
Handoff signaling cost is the number of messages sent over the network during handoff processes. The smaller signaling cost the better performance of the proposed scheme. The size of data or any additional packets can be represented as S name . The MBMA signaling cost (8), the producer sent a packet to update the server about new name prefix, a consumer sent a packet to query the server to obtain a forwarding hint. Then the consumer can forward the Interest through optimal path to the producer and received data via the same route. The packets delivery cost is presented in (9).
The LBMA signaling cost (10) is formulated when producer sent a packet to update the new and old AR about new domain prefix, a consumer sent an Interest towards the old AR. Then old AR forwards the Interest with the location to the new AR in triangular passion, the data is also routed back to the consumer via the same route. The packets delivery cost is presented in (11).
The CDBMA signaling cost (12) is formed in a similar passion with MBMA, it also provides optimal [1] packet delivery path. The only difference is the server can update the entire FIB of the ARs within the domain. The packets delivery cost is presented in (13).
The IBMA signaling cost (14) is formulated in the same manner with LBMA, the only difference is an encapsulation of Interest packets between old and new AR. The packets delivery cost is presented in (15) which is also triangular routing path.
The PMSS signaling cost (16) is formed when the producer sent a mobility packet to update the anchor, the anchor broadcast the mobility Interest to update the intermediate routers. A consumer can send an Interest and received data via the optimal route. The packets delivery cost is presented in (17).

IMPLEMENTATION 5.1. Model design and implementation
The formulated models for handoff latency and signaling cost analysis is designed and implemented using Python programming in the Spyder IDE environment. The Spyder IDE is one of powerful development tool for python language that provides IPython console with an interactive prompt to execute codes. The analytical model was verified at the implementation phase to ensure the correctness and accuracy of programming codes ready for implementation design. The model was validated using parameters stated in Table 1, to observe and validate the accuracy of the model's applicability related to producer mobility problems such as handoff latency and cost, for all the driven analytical models of existing and proposed schemes.

RESULT ANALYSIS AND DISCUSSION
For the numerical result analysis for both handoff latency and signaling cost, we set a=1, b=c=5 and d=9 as used from the literature [16,19], 16 bytes are used for additional field upon normal Interest packets as presented in Table 1. The numerical result on the handoff latency analysis have shown in Figure 3 (a), (b) and Figure 4 (a) that illustrate the effect of varying different transmission cost between old and new AR, ARs and server. In addition, compare the handoff latency of MBMA, IBMA, LBMA, CDBMA, and PMSS. In Figure 3 (a), the latency of IBMA and LBMA significantly increase by varying b, because a packet must route through old AR was also known as a home router. In Figure 3 (b), the latency of all approaches was increased significantly, due to the effect of varying cost between ARs that can directly affect any approach. In Figure 4 (a), the latency of MBMA and CDBMA significantly increase by varying d, and linearly for IBMA, LBMA, and PMSS as result of querying routing update from the server, control serves or resolution handler. Hence, we can realize that the handoff latency of propose scheme (PMSS) is the lowest with the effect of varying b, c, and d in the midst of five different approaches. The numerical result on the handoff signaling cost and optimal data delivery analysis have shown in Figure 4 (b), Figure 5 (a) and (b). The handoff latency and signaling cost are related based on the hop count analysis; there is certainty that when latency is high the signaling cost will also be high. From Figure 4 (b) we can see that the signaling cost of IBMA and LBMA are significantly increased, while for MBMA, CDBMA is uniform due to the increase of b that directly affect the transmission between old and new AR. Moreover, Figure 5 (b) shows the signaling coat of all approaches increases significantly, due to the effect of varying c. In Figure 5 (b) the variation of d affected only MBMA and CDBMA. Therefore, we can realize that the proposed scheme has the lowest signaling cost among five different approaches. shows the optimal packets delivery for our proposed PMSS. In Figure 6 (a), the packets delivery of LBMA and IBMA are significantly increased, while that of MBMA, CDBMA, and PMSS are uniforms. Since from the literature, the approaches that use DNS-like server or control server transmitted their packages via optimal route after the handoff. Therefore, PMSS is optimal as it goes with MBMA and CDBMA. In Figure 6 (b), the variance of c affected all the approaches, hence PMSS and MBMA and CDBMA have the lowest cost. Therefore, we can conclude that PMSS is optimal in terms of packets delivery path.

CONCLUSION AND FUTURE WORK
The insufficient research on data path optimization after handoff causes serious degrading of seamless mobility support in NDN. In this paper, we have proposed a producer mobility support scheme that minimizes the handoff latency, signaling cost to reduce the unnecessary Interest packets loss. We introduce mobility Interest, broadcasting strategy and modified immobile anchor router to perform a dual function, that is anchoring and broadcasting. The numerical result shows that he PMSS reduces handoff latency, signaling overhead cost and improve data packets delivery via optimal path once a content producer relocates to another PoA compared to existing approach. Predominantly, handoff problems can be investigated using the analytical method, whereas other issues like the determination of packets delivery, packets losses, scalability and path optimization needs to be conducted through simulation investigation for a better result. Henceforth, in our future work, we will focus on simulation investigation would be carried out for the overall evaluation of handoff performance, path optimization, and packets delivery compared to specific existing schemes.