Medium Access Control Layer for Internet of Things Edge-Side Network Using Carrier-Sense Multiple Access Protocol ^†

Abstract

The Internet of Things (IoT) has recently received a great deal of research interest due to its broad range of applications. One of the important layers in IoT applications is known as edge computing where resource-constrained devices at the edge form a simple type of network to sense required data. A more powerful edge device is responsible for collecting all sensed data to be transferred to the upper layers. A critical focus is therefore placed on maximum rate of data collection, requiring effective and intelligent solutions to coordinate the channel access of the devices. Medium access control (MAC) protocols take this responsibility as their design mission. Carrier-sense multiple access (CSMA) has been a baseline MAC scheme and many previous traditional networks utilized a CSMA-based solution. The motivation of this paper is to study the performance of a typical network at the edge through the CSMA theme. A practical network is constructed to assess the channel throughput performance via a commercially available radio transceiver. The practical performance observations indicate the suitability of the proposed CSMA-based solution.

1. Introduction

Wireless communication and networking have long been applied to provide more intelligent solutions in diverse areas of engineering. A typical networking-oriented application aims at creating an environment for devices with the goal of sharing knowledge in a wireless manner. Devices, often referred to as objects, work together to accomplish a common task, such as detecting a wildfire with a fleet of unmanned aerial vehicles (UAVs) [1]. In recent years, the developments and innovations in terms of producing more powerful objects have resulted in an increasing number of objects. This also extends the application spectrum to a wide range, from terrestrial to underwater and aerial. Objects share the same transmission medium which makes the coordination of devices a critical concern, in order to avoid the inefficient utilization of capacity. Many researchers and practitioners in this area have placed an important focus on the advancement of efficient solutions to ensure a high level of quality of service [2].

The Internet of Things (IoT) has been a popular technology that permits physical objects to interconnect with each other with no human control [3]. It has been a central point in the fourth industrial revolution (Industry 4.0), and after this, IoT has been utilized in all aspects of life. To better understand the complete nature of IoT, a well-known architecture including main layers is presented in Figure 1. The physical layer at the bottom is usually located at the edge of the network. Therefore, this layer is responsible for collecting real-world information through a set of sensing equipment. In this layer, traditional wireless sensor network (WSN) nodes (small low-cost devices) are mostly used to form a simple network [4]. This essentially renames the WSN nodes as IoT nodes. Since IoT nodes must contend for the transmission channel with the same probability, it is of high importance to develop medium access control (MAC) approaches to manage the channel access strategy of the IoT nodes.

An efficient MAC protocol should be able to avoid simultaneous transmissions from multiple IoT nodes. Otherwise, a packet collision will occur at the receiver side, resulting in the destruction of all collided packets. Several MAC solutions for IoT applications have been proposed to achieve a perpetual operation among the nodes [5]. Most of the proposed protocols employ the theme of a classic approach, carrier-sense multiple access (CSMA). A common drawback of the protocols are the unrealistic performance evaluations, as simulation tools are used to assess performance [6,7]. These models make a great number of assumptions and ignore the problems to be faced in practice. This paper has its basics on the practical implementation of the CSMA protocol using a real-world transceiver. A small-scale single-hop topology is created to observe the performance of the CSMA protocol in terms of channel throughput, representing the successful ratio of channel usage. In addition to CSMA, we also run the experiments for the ALOHA protocol which is perhaps the first MAC approach in history. The performance results indicate interesting outcomes with respect to environmental impacts. In the following sections, we will first start with a brief background and definitions of models. Then, the performance outputs obtained from the deployed test-bed will be presented in association with varying channel loads. An explicit conclusion along with possible future work directions is supplied to conclude the paper.

2. Background with System Models

2.1. Fundamentals of ALOHA and CSMA

In the literature, the ALOHA protocol was the first MAC scheme with its key advantage of simplicity. ALOHA enables objects to start transmission immediately when a packet is generated to be sent. This causes a lot of packet collisions, degrading the overall throughput and energy performances. The collided packets are considered to be lost and a unique re-transmission policy is operated for each collided packet. The maximum number of re-transmissions for a packet is assigned to 7, so that, after this limit, a collided packet is completely discarded from memory. ALOHA assumes that each object has the same capabilities and allocates the same channel access opportunity. The throughput of the ALOHA protocol can be analytically computed as:

T = Ge^−2G

(1)

where G represents the offered channel traffic ratio with the main assumption of an infinite number of objects. The throughput curve of the ALOHA protocol is plotted in Figure 2. It can be clearly noticed that the maximum throughput value of 0.5 can be achieved at a traffic load of nearly 0.18.

CSMA has been proposed as an alternative method to handle the deficiency of the ALOHA protocol, allowing the objects to sense the transmission medium before starting to send a packet [8]. Therefore, an object checks the medium to make sure that there is no ongoing transmission. This requires special equipment to have the capability of sensing in practice. There have been several variants of CSMA for different purposes, such as detecting collisions and collision avoidance mechanisms. This study considers the 1-persistent CSMA version in which an object performs a continuous sensing process until it finds the medium as collision-free or idle. This solid strategy has a strong mechanism in reducing the likelihood of collisions, but there is still a possibility of collisions if more than one object senses the channel idle at the same time. The throughput expression is given below. Here, the parameter α denotes the proportion of propagation delay to packet transmission time which has a great impact on the throughput (Figure 3).

$$T={\displaystyle \frac{G{e}^{-G(1+2\alpha )}[1+G+\alpha G(1+G+\frac{\alpha G}{2}\left)\right]}{G\left(1+2\alpha \right)-\left(1-e^{-\alpha G}\right)+(1+\alpha G)e^{-G(1+\alpha )}}}$$

(2)

2.2. Hardware with Transceiver

A low-cost simple object is created which includes an Arduino UNO platform as the processor and an NRF905 Transceiver module as the wireless communication unit [9]. A view of the NRF905 module can be seen in Figure 4. This module provides a communication range of up to 250 m in line of sight. It has variable data rate options which are 250 KBps, 1 MBps and 2 MBps. The maximum payload size of a packet is 32-byte. Extra bytes for address and CRC can be appended.

2.3. Overview of Single-Hop Topology

We created a single-hop topology with 6 transmitting objects and a receiver is deployed at the center, as shown in Figure 5. This small-scale topology is located in an office environment and the distance between the objects is carefully arranged. Each object transmits at the same transmission power and follows the Poisson traffic model for channel load. It is highly believed that this type of small network would give a strong sense in assessing the performance. Therefore, it is worth mentioning that deploying a large-scale network may present a slightly different performance. We run the network several times and an average value of these trials is recorded.

3. Performance Results

This section presents the throughput curves of both the ALOHA and CSMA protocols with respect to varying channel loads. Figure 6 demonstrates the throughput performances where each value is an average of several runs to make a fair assessment. When comparing the throughput of ALOHA with the analytical results presented in the previous section, we see that there is a significant difference. The maximum achievable throughput increases to around 0.33, which was limited to almost 0.18. The reason behind this increment was successfully defined in our previous study [10]. The throughput has a significant enhancement due to capture effect phenomena, allowing the first-arriving packet to be received in a collision at some extents. The results of the capture effect on ALOHA throughput using a different practical transceiver module exhibits a good match over the results presented in this paper. Therefore, a main finding of this study is to remind the researchers in this area to take the capture effect into consideration when developing a practical scenario.

The throughput of CSMA, on the other hand, is higher than ALOHA in all traffic loads, confirming the superiority of CSMA over ALOHA. The throughput curve has an increasing trend, reaching its maximum value of approximately 0.42 at the highest channel load. As explained in the previous part, the throughput of CSMA depends mostly on parameter α. This parameter is a dynamic one depending on the characteristics of both radio transceiver and environment. The results presented prove that the practical throughput can match the analytical values at low traffic loads. Also, the capture effect should have a small impact on the throughput as packet collisions are still likely to occur. To sum up, this paper reveals the effects of real-world conditions on the performance of two well-known MAC protocols. It encourages researchers to consider these cases when designing new protocols in future.

4. Conclusions

This study revisits the two popular MAC schemes, ALOHA and CSMA, for random-access wireless networks. The performance of these two baselines in terms of channel throughput in a practical scenario is examined. A low-cost transceiver is used to observe the throughput performances in a small-scale single-hop topology. The throughput performance of ALOHA is significantly improved in practice, thanks to the capture effect. On the other hand, CSMA has a similar throughput trend as in the analytical model. Therefore, when designing novel MAC protocols, this paper hopefully provides practical cases, instead of simulation-based efforts.