Dr. Shintaro Mori’s official website

Ⅲ.A. Development of an efficient caching scheme for WSNs based on ICN

This study aims to introduce an ICN architecture into WSN systems, not wired network systems. In the caching method in ICN, the system should establish a link between the node that requests the contents and the node that has them. Here, there are two types of caching methods: on-path caching, where the relays on the route hold the cached data, and off-path caching, where other nodes store the data on their local storage. In general, in wireless communication, the nodes near the transmitting and receiving nodes have specific characteristics, which can be obtained overheard by communication via radio signals. In this study, we develop a novel scheme for off-path caching in WSNs based on the feature of overhearing. Fundamentally, ICN architecture has been designed for wired communication systems, and off-path caching techniques have not been introduced or have been implemented using separate channels. In addition, the proposed scheme has the advantage of off-path caching without overhead and without introducing any new complicated mechanisms.

Ⅲ.B. Cross-layer optimization and computer-simulation evaluation

In order to effectively transfer caching data from the sensor node, we will develop a new wireless communication technology using the successive interference cancellation (SIC) technique. The SIC technique has been developed as an elemental technology for the cellular-network field. It has been implemented and commercialized as an LSI chip for a next-generation high-speed mobile communication system. In the SIC technique procedure, the received signal is restored according to the strongest signal strength. If the decoding is successful, the decoded data is re-encoded and then subtracted from the received signal. Through repeating this operation, the strongest interference can be removed, which helps to improve the probability of successful decoding. In contrast, the concept of the SIC technique has not been considered in the network field, and this study emphasizes this factor as a novelty.

Ⅲ.C. Development of a secure cashing scheme using blockchain and its initial evaluation

Here, we conduct theoretical analysis using a statistical model, assuming it is used in WSNs. In addition, we will discuss the feasibility of protocol design with researchers in both the wired and wireless network fields. ICN architecture is relatively innovative, and it is difficult to evaluate using hardware-based experiments. Therefore, for the evaluation of the proposed scheme, we will use not only theoretical calculations but also computer simulations. Note that in this evaluation, we implement the simulator using C++. At the same time, we have recognized the necessity of a mechanism for secure sensing-data sharing within the WSN system, prior to evaluation based on testbeds. Therefore, to establish this mechanism, we introduce blockchain technology and adopt it into the development system. In particular, it is necessary to achieve a distributed data management approach that differs from conventional host-oriented networks, such as IP networks, and blockchain has significant suitability for WSNs based on ICN.

Ⅳ.A. Literature review regarding efficient caching scheme

In the wired network field, ICN technology is attracting attention as a next-generation Internet architecture, gradually spreading to wireless networks. For example, some proposals have been investigated in cellular networks, such as caching in base stations, gateways, and routers. In other words, it enables the reduction of backhaul network traffic thanks to the distribution of the same data. In this situation, ICN is similar to CDN, but the difference is that CDN caches contents within the network server, whereas ICN stores contents widely in the terminal nodes.

Based on an analysis of a large number of ICN-related studies, the cashing method was classified into two types: on-path cashing and off-path cashing. Namely, on-path caching stores content in the relay nodes on the route, whereas off-path caching actively copies content to nodes outside the route. In a typical ICN framework, on-path caching is provided as standard in DONA and NDN. There are four off-path caching methods: user cooperative caching methods, data selection methods, and dynamic caching methods for video streaming delivery with guaranteed quality of service (QoS).

Ⅳ.B. Network model

In ICN-based WSNs, the content data are stored in SNs. Similar to CDNs, ICNs do not distinguish between original data and copied data. As shown in Fig. 1, in the proposed scheme, the sensing data are stored in the cache memory in the SN as cached content data. Regarding the procedure for retrieving any sensing data, in typical WSNs, the SN connects to the cloud server via the Internet, where the sensing data is centrally managed. Therefore, the user who desires to obtain sensing data only needs to access the cloud server. On the other hand, in the proposed scheme, the user accesses the gateway node that belongs to the WSN because it is necessary to find the relevant sensing data from the content data stored in the network. The gateway node, acting as a subscriber, broadcasts a request to acquire sensing data within the network and launches a procedure to conduct the investigation. When the sensor node that holds the desired sensing data from the user responds to the previous request and transmits the relevant sensing data, it completes the task.

As a mechanism to enable routing data from the publisher to the subscriber, the proposed scheme makes it possible by recording the tracking information of the request packet sequentially when it is flooding. In addition, since the tracking information is contained in the header of the request packet, it is possible to distinguish among each request, and it is also possible to identify the next destination (sender) to be sent.

Ⅳ.C. Key concept of the proposed scheme

As shown in Figure 2, the proposed scheme uses both on-path and off-path caching. Regarding on-path caching, the scheme is implemented in the same manner as other studies, where sensing data are forwarded from the publisher to the subscriber in the response process, and the relay node stores the sensing data in its own cache memory. On the other hand, in the case of off-path cashing, based on the overhearing phenomenon, the relay node accumulates the overheard sensing data, which is sent from the SNs that are in the relay node’s communication coverage area.

Due to the overhearing phenomenon, the proposed scheme does not require any unnecessary wireless communication between the relay node and the adjacent sensor node. In contrast, to perform overhearing-based communication, it is necessary to consume more power for radio signal processing and further increase the cache memory capacity. Even if the energy consumption of the microcontroller and other processing in the SN increases, the total energy consumption of the SN can be improved by reducing the number of wireless communications by introducing the proposed scheme because the energy consumption of the wireless communications is smaller than that of the calculations. In addition, the cost of extra cache memory is more than paid for by the savings in radio frequency resources, which can reduce the number of wireless communications. However, since the capacity of the cache memory in the SN is not unlimited, the rules for discarding sensing data stored in the cache memory must be decided according to the environment in which the proposed method is adopted. This question is not covered in this paper, and the detailed decision-making process is a future work.

Ⅳ.D. SIC-based cashing scheme

SN can receive signals from adjacent SNs based on the overhearing phenomenon, and it can be used positively on the basis of the aforementioned method, but it can also be a double-edged sword that causes interference that reduces the data reception probability. Therefore, we use the SIC technique to reduce interference from other SNs in off-path caching based on the received signal. As shown in Fig. 3, let $P_i$, $j$ denote the signal strength transmitted from the $i$-th SN to the $j$-th SN, the set of SNs that the ${ℳ}_j$-th node can receive based on the overhearing phenomenon. Letting $P_i$ denotes the desired signal, in general, conditions under which the receiver can decode is given by

where $\Lambda$ is a threshold value that can be successfully decoded, ${\sigma }^2$ is an ambient noise, respectively.

In Fig. 3, $y$ denotes the $j$-th SN’s received signal, which is given by

Figure 4 shows the procedure of the proposed decoder. In the first decoding process, the signal with the strongest received signal is decoded. When the signal can be decoded correctly, the restored signal is encoded again, and the signal is subtracted from the received signal. In the same manner, it is repeated until no more signals can be decoded.

Ⅴ.A. Protocol stack of the proposed scheme

For an ICN-based WSN system, in the case of introducing the overhearing phenomenon and caching method based on the SIC technique, the protocol design should be flexible and integrated. A cross-layer design-based approach is often suggested for optimizing the protocol stack. Cross-layer design is a technique that achieves a flexible layer design by sharing information between layers, unlike conventional layer designs (such as the OSI reference model). In particular, the design concept is based on the interconnection between adjacent layers and the protocol stack, as shown in Fig. 5. In the proposed scheme, the TCP/UDP, IP, and MAC layers of the conventional architecture are redefined as the ICN layer. The ICN layer is then divided into two types of control and data transmission functions, called the C-plane and U-plane, to achieve each function.

Ⅴ.B. Procedure of the cross-layer-design signal processing

Figure 6 shows the signal processing procedure for cross-layer design in the proposed scheme. The sensing data acquired from the observation area are stored in its own cache memory. As well as storing its sensing data, the cache memory also stores as much as possible in other SNs that it has overheard. The TX/RX RF (radio frequency) module sends and receives sensing data. On the other hand, the microcontroller not only manages the sensing data and routing table but also controls the cache memory and RF module. At this time, the signal transmission required for this control is designed according to a cross-layer design based on side information from the upper layer to the lower layer.

Figure 7 shows the detailed signal processing procedure in the RF module and wireless communication channel emulator, as shown in Fig. 6. As shown in Fig. 7(a), the transmission packet is encoded with a cyclic redundancy check (CRC) for error detection, and then each bit is modulated using the binary phase shift keying (BPSK) method. On the other hand, as shown in Fig. 7(c), the receiver performs soft decoding and then applies SIC technology using the decoded replica. The decoder and encoder in Fig. 7(c) are the same as the blocks shown in Fig. 4, and they perform packet restoration and signal subtraction processing according to the procedure described above.

Ⅴ.C. Simulation model (wireless propagation model)

As shown in Fig. 7(b), the packet error probability is calculated based on the received signal strength indicator (RSSI). In general, the link budget can be expressed by

where letting $P_{\mathrm{TX}}$ and $P_{\mathrm{RX}}$ denote the transmitter and receiver power, letting $G_{\mathrm{TX}}$ and $G_{\mathrm{RX}}$ denote antenna gain in TX and RX, and $L_{\mathrm{TX}}$ and $L_{\mathrm{RX}}$ denote the attenuation in the TX and RX electronic circuits, respectively.

In the above equation, $L_p$ is the path loss of the wireless channel. The packet error probability required for computer simulations is calculated based on the signal-to-noise ratio (SNR) and bit error probability. Specifically, the received SN signal power, and the SNR $\gamma$, is given by

where $K_B$（= $4.0\times {10}^{-21}$ W/Hz）is Boltzmann’s constant, ${\tau }_o$ is the system temperature of the hardware device. We can theoretically calculate the bit error probability $p_b$ of BPSK under Rayleigh fading channel.

Ⅴ.D. Computer simulation

A computer simulator was implemented using the C++ language for the RF module, radio propagation model, and cache memory, as shown in Figs. 6 and 7. As a simulation scenario, SNs were randomly deployed in the observation field (i.e., the positions of the SNs were determined based on random uniform distribution). In this case, the combination of publisher and subscriber was randomly determined, and the shortest path was selected between the two nodes. In the simulator, the routing path was calculated based on the Dijkstra algorithm. The constant parameter settings of the RF device were determined based on the XBee module, which is widely used in wireless communication in WSNs.

Figure 8 shows the percentage of successful caching for all SNs $\rho$. In accordance with the effect of the overhearing phenomenon, $\rho$ increases as the number of SNs increases, but $\rho$ decreases after reaching a maximum of 180 SNs. In comparison with the method that does not use the proposed cashing mechanism, the number of SNs was improved by 1.42 times, 2.61 times, 8.85 times, and 17.0 times when the number of SNs was 100, 180, 600, and 1,000, respectively.

Figure 9 shows the average number of multi-hops when the request packet is correctly transmitted from the publisher to the subscriber. As the number of SNs increases, the number of relay nodes that can forward packets also increases; thus, the average number of hops also increases. The proposed scheme uses significantly more cached data than the comparison scheme (Fig. 8), and the probability of a request packet hitting a cache is also higher, improving the average number of multi-hops. When the number of SNs was 50, 100, 150, and 200, we obtained improvements of 3.35%, 8.51%, 10.8%, and 10.9%, respectively.

Figure 10 shows the average transmission distance versus the number of SNs for the transmission link between the publisher and the subscriber. In the region where the number of SNs is less than 100, the proposed scheme is not effective because of insufficient caching, but in other regions, the performance is improved. Specifically, the system can improve by 6.46%, 10.5%, 12.3%, and 12.2% for the case number of SN are 50, 100, 150, and 200, respectively.

Ⅵ. Development of Secure Caching Scheme using Blockchain