Edge analytics is the latest technology in the data science sphere which aids in collecting and automatically analyzing data at the point of network devices rather than transmitting it to a centralized data center. Currently, most of the industries including transportation and logistics implement Internet-of-Things (IoT) devices at every place where their operations function. These devices generate an abundance of operational data that are highly complicated and difficult to manage. Traditionally, such huge data is first transmitted to cloud-based data centers for storage over the network. It is then processed and analyzed to bring out insights utilizing conventional analytical tools. Subsequently, the resulting insights are transmitted back to the operations area for responsible stakeholders to act accordingly. This whole process is time-intensive and dependent hugely on network traffic performance. Often it takes more than usual, and pro-active action cannot be taken, causing a huge loss. With Edge analytics, it removes the requirement to transmit entire data generated from embedded sensors to the data center for analytics. Insights can be produced immediately as the data is generated and essential actions can be taken to avoid any future incidents. It, therefore, helps in reducing the latency in decision making by detecting on-premise problems in real-time.
Shortcomings of existing Performance Management System in Transportation and Logistics: As per a survey conducted by InformationWeek, almost 91% of transportation and logistics organizations have implemented the business intelligence and data analytical solutions. However, merely 46% of those solutions are projected to elicit efficient action-oriented insights. The below table highlights some of the existing offerings of such solutions and their related limitations.
Figure 1: Offerings and limitations
In the subsequent sections of this article, an analysis has been put-forth on how Edge analytics can overcome the above-mentioned limitations.
Prospects in Transportation and Logistics: Transportation and logistics industry need to perform with speed, accuracy, and security, thus need a solution that can extract information and provide advanced analytics around these three crucial parameters in real-time. Like any other industry, the transportation and logistics industry also operate on millions of sensors, devices, and computing gadgets. These devices result in the accumulation of humongous data and according to an article by Sciforce, 73% of such accumulated data is never used. To beat the highly competitive market and build strong customer experience, it is getting critical for organizations to monetize this untapped data by keeping themselves abreast of the latest technologies and align their business goals accordingly. To add to it, encompassing analytical capabilities is one of the most essential ingredients for an organization to gain a competitive advantage and hold a strong market position. As per the latest trends, the transportation and logistics industry is also being revolutionized with the advent of autonomous vehicles. The global autonomous vehicle market is predicted to grow up to $590 billion by 2026. These vehicles have various sensors embedded like digital cameras, radars, and lasers that are used to communicate with other vehicles on the road to avoid a collision. Additionally, these devices are also used to understand weather and road conditions. All the information is then fed to sophisticated algorithms and models to make decisions. These decisions may range from how the vehicle should act when it encounters a traffic jam, construction areas, lane change, or any other dynamic situation. Any latency in such decisions may cause havoc on the road and put lives in danger. Here, Edge analytics comes to a rescue that allows these algorithms and models to run nearby the devices and make tactical decisions right on the spot where the action is required to be taken. This is just one aspect of Edge analytics in transportation and logistics, however, there are many more yet to be researched and discovered.
Workflow for Edge analytics Approach:
In general, an Edge analytics architecture consists of three layers: Edge Sensors and Actuators, Edge Gateways, and Edge Devices. The Edge Sensors and Actuators sense data and collect it from various IoT devices. Then, Edge Gateways with continuous power-supply, cleanse, and aggregate the data. Lastly, Edge Devices run the preliminary analysis and if any action is required to be taken, a decision in real-time is triggered. Further, only relevant data is sent to data centers for deeper analysis and strategic purposes. This workflow not only avoids delays in critical scenarios but also reduces the cost of network bandwidth.
Figure 2: Edge analytics architecture
Vendor Landscape for Edge Analytics
The current market has many leaders who are empowering the Edge analytics solutions; however, the prominent ones include Cisco, International Business Machines (IBM), Hewlett Packard Enterprise (HPE), and Dell. IBM developed Watson IoT Platform Edge analytics to embrace the idea of analysis at the Edge of IoT devices. It has done partnership with other vendors to integrate its legacy of cognitive analytics with its vendor’s Edge analytics devices. For instance, IMB has collaborated with Honeywell BuildingSense to avail Edge analytics in retail stores. Likewise, IBM partnered with Cisco’s Industrial IoT Analytics platform and performs device monitoring, equipment health measuring for preventive maintenance, and other manufacturing-related tasks. IBM even has published an online Watson Edge analytics Cookbook which guides developers to customize the solution and provide the required technical specifications along with instructions to deploy the solution. Transportation and logistics stakeholders can take advantage of this cookbook to implement the relevant and customized Edge analytics solutions in their organizations. Furthermore, HPE and Dell: major players in computer hardware, also have come together to build models for Edge analytics. HPE has three lines of Edge-oriented servers.: Edgeline EL20, EL1000, and EL40000. They have so far exhibited their applications in manufacturing industries, oil and gas drilling, and smart city development. Dell has four lines of Edge-oriented servers: Gateway 3001, 3002, 3002, and 5000. Gateways 3001 has shown potentials in refrigeration monitoring and agricultural applications. Gateway 3002 and 3003 are tailored specifically for the transportation and logistics industry. They can be used for managing transportation assets, traffic monitoring, and rail operations management. The Gateway 5000 is the advanced version of 3000 series and provides deeper analytics.
Current Industry landscape
The necessity to adapt Edge analytics is increasing every day across industries. As per a survey report, the Edge analytics usage is going to increase up to USD 7.96 billion by 2021. The below figure suggests that implementing it is in progress for almost 57% of industries and as per the projections, many more would join the force soon. We can also notice from the figure that organizations have already initiated the thought process to consider the potentiality of Edge analytics and none have responded that it is not in their organizational plans which bolsters the utility of Edge analytics.
Figure 3: Edge analytics implementation status in various organizations
Potential Use-Cases in Transportation and Logistics
Transportation and logistics are an integral part of most businesses and provides a strong foundation for the economy. It is going through many transformations and looking for sustainable and smart solutions. This objective builds a strong base to invest in advanced technologies and IoT infrastructure is one such technology that has greater prospects in this industry. As per the statistics provided by Statista, transportation and logistics owe up to $40 billion of investment on IoT devices and services till 2020. These IoT devices would, in turn, generate an enormous amount of data and the corresponding insight can revolutionize the business model and bring in a gamut of opportunities. Below are some of the use-cases of Edge analytics in the transportation and logistics industry.
Device-Failure: The data ingested from 3D-sensing cameras, and other in-built sensors in devices can further be analyzed, visualized, and integrated with static data such as service schedules, previous damages and repairs to provide the current condition of equipment and vehicles in real-time. The detection of patterns in anomalies and performance can help in inventing future failures and thus, can save costs on maintenance and reduce production downtime. The below figure exhibits how patterns in the vibratory displacement of the embedded devices like gears, accelerometers, etc. are analyzed over a while, and then a point of maintenance is projected before the machine reaches the point of failure.
Figure 4: Predictive Maintenance by Edge Analytics
Cargo-Temperature Monitoring: One of the key challenges specifically with logistics is that the perishable items often get ruined due to unfavorable climatic conditions of varied locations it passes through. So, sensors are mounted in the cargo vehicles that continuously monitor the temperature and humidity level. With Edge analytics devices, the cargo temperature can be regulated automatically based on the relevant location weather which alleviates the food spoiling.
Safety: Human resources are an integral part of the transportation and logistics industry. Edge analytics not only safeguard vehicles but also the driver who is driving them. The videos collected from in-built cameras are critically analyzed to explore patterns of fatigue and distraction in the drivers’ behavior. Utilizing embedded Edge analytics, the vehicle itself can alert the driver in real-time to take rest or pull-over, thus, ensuring their safety on the road.
Hesitance to invest in newer technologies: Edge analytics is still in its early stages. There is a very limited set of vendors who are providing Edge analytics solutions in the market. Also, the presence of a wide range of distinct solutions leads to indecisiveness. Additionally, in the transportation and logistics industry, the workforce is not generally digitally skilled. This gap can only be filled with a considerable amount of training and infrastructure and thus, the organizations need to exclusively invest to make it successful.
Computationally intensive: Edge analytics is a growing technology; thus, the market lacks enough hardware products that are capable of storing such a huge amount of data on-premises and perform complicated analytics. Moreover, the implementation process requires draconian efforts. Organizations need to interconnect all the IoT devices. The collected data would then need to be analyzed, and analytical models are required to be trained. Subsequently, the integrated data will have to be fed into the operating and information systems. All these processes are time-intensive and demand high-end equipment.
Lack of Universal Standards: Currently most of the industries lack well-defined and stringent standards from the perspective of data. However, the logistics sphere has taken the initiative and the Digital Container Shipping Association (DCSA) aims to design standards for data processes. Other organizations still need to buckle-up if they want to leverage their data insights to earn benefits.
As per International Data Corporation (IDC), a market research firm, approximately 45% of data will be managed and analyzed at the Edge of devices. So, the transportation and logistics firms with 100% total cost of data operations on the cloud can save at least up to 60% of the cost. Overall, Edge analytics reduces the maintenance cost tremendously and provides an optimum, secure, and reliable system for faster decision making.
Data redundancy: Oftentimes, data generated is noisy and contains irrelevant data points. Transmitting this load of data eats up large bandwidth and costs a ton. Edge analytics offers storage-optimization algorithms that retain unique data points for different attributes. Since we cannot completely ignore the transmission of data to the central data location, only the most relevant data would be transmitted without overloading the network. Furthermore, it allows to define a customized storage retention period, and after which the data is archived. This, in turn, avoids any bottleneck situation and saves costs.
Network Traffic: Since, Edge analytics filters and sends only the essential data to the centralized data center, it prevents network congestion and in turn, saves the cost of expensive bandwidth. This factor can be corroborated by the claim made by the NEC Laboratories (previously known as Nippon Electric). They mentioned that using their device viz., Geelytics can reduce the bandwidth cost by 99%.
Scalability: Numerous sensors embedded in the factories, warehouses, autonomous vehicles, and 3D-sensing cameras generate a variety of data in various forms like text, videos, images, audios, etc. Storing, managing, and eventually integrating such heterogeneity in data is a daunting task. Edge analytics architecture does all these tasks upfront at the device level itself and thus, reduces the computational load on data centers. This further ensures an efficient division of labor and many more IoT devices can further be deployed allowing data to scale reliably.
Equipment improvement: Analytics embedded into the Edge of IoT devices can help in recognizing the patterns and predicts potential failure. It can trigger real-time alerts for outages. Moreover, automated self-correcting measures can be triggered. Thus, an overall downtime can eventually be reduced.
Data Security: Cybersecurity is the greatest concern while data is transmitted to the centralized data center. But, with the application of Edge analytics, since data stays within the firewalls, it cannot be intercepted. Data is thus made hack-proof and any breaches would be impossible.
Define Data Strategy: To exclusively extract business benefits, enterprises need to define a well-rounded data strategy and align their business strategies accordingly. The factors that pivot the act to define a data strategy encompass heterogeneity in data and humongous data management. The data strategy should have procedures to harness data to act intelligently and provide real-time contextual information. The data strategy should focus not to eliminate the centralized data analysis but should complement the existing analytical solutions. Additionally, it should include well-defined specifications for the capacity of Edge devices to establish an optimized infrastructure.
Identify vendor: Since data is the key asset, the vendors for Edge analytics should be chosen keenly. A vendor should be evaluated first based on certain criteria. For instance, enterprises who have recently ventured into employing Edge analytics should consider a vendor who can avail an end-to-end Edge analytics services rather than just delivering the end product needing to customize as per the requirements. Additionally, the vendors should have a definite implementation plan and an extensive data set of performance measurements with an emphasis on accuracy. Most importantly, the rules and policies should always maintain transparency and ensure the integrity of the data.
Edge analytics devices installation: Edge analytics solutions require an extensive amount of time to get deployed. Thus, an incremental approach to deployment is recommended. Moreover, organizations defer in terms of their objectives, operations, and priorities, so the initial phase should be dedicated thoroughly on identifying, evaluating, and prioritizing essential use-cases that can provide faster profits and that are easy to justify the investments. Alongside, experts and other stakeholders are required to spot key performance indicators (KPIs). In the next phase, the actual installation of Edge analytics tools should be done and tailored according to the unique environment. Post the implementation, the performances should be measured based on the previously identified KPIs and if the target is met, the next use-case should be processed. To add to this, an extensive study needs to be done on the existing software and hardware systems. Since the computing power and processing would be high-end, will have different specifications and run on varied operating systems, the integration with the existing system would be a challenge. If the study suggests an upgrade of existing systems, the project management team should include the process in their schedule and a diligent plan should be laid out to successfully sail the process.
A Collaborative effort by stakeholders: The entire infrastructure of Edge analytics cannot be established just with software and hardware engineers. Other personnel such as people with domain and device knowledge are essentially crucial to set-up the process. So, apart from data experts, field engineers, network architects, and application engineers would be required to work together. Since, Edge analytics technology is routinely going to be managed by on-site managers, drivers, and other end-user workers, extensive training should be designed to make them aware of its utility. Moreover, since solutions evolve, knowledge retention is important. A repository to record knowledge and insights should be created that can base future scalability and improvements.
Acquiring start-ups: Data science has evolved recently as a prominent field. There are rapidly growing start-ups with unique solutions and innovations who are working actively on analytics. In the long-run, organizations should be on a look-out to acquire start-ups to develop an in-house eco-system of analytical solutions.
The below figure summarizes the above implementation strategies:
Figure 5: Implementation Strategies
A proposed outline of the Edge Analytics deployment steps is as below:
Figure 6: Proposed Edge Analytics deployment steps
- Data strategy: This process would need thorough research, frequent meetings, and collaboration with management and other stakeholders.
- Cost-benefit analysis: In this process, the cost of investment and comparison with the expected benefits are analyzed.
- Identify vendor: In this process, a suitable vendor would be searched, and contract signing is finalized.
- Compatibility evaluation: This process is required to examine the existing system’s capacity, its compatibility with the latest technology, and the need to upgrade the existing devices.
- Define KPIs: In this process, the managers along with vendors would define the key performance indicators to measure the system performance.
- Prioritize use-cases: In this process, the potential use-cases are identified and prioritized.
- Deployment: This process is time-intensive, and the period of deployment may vary depending on the use-case being implemented.
- Performance measurement: Post the deployment, in this process, the performance of devices would be measured to ensure reliability, accuracy, and address any fallacies.
Edge analytics enables industries to capitalize on data that often remained unnoticed. Transportation and logistics industry which relies on speed and accuracy is getting laden with Internet-of-Things (IoT) devices. Pushing analytics algorithms to IoT devices would diminish the strain on centralized systems and allow them to take efficient reflex actions. The prominent vendors who are availing the Edge analytics solutions are Cisco, International Business Machines (IBM), Hewlett Packard Enterprise (HPE), and Dell. Their products have exhibited potential in various industries.
Potential Use-Cases: Edge analytics is essential in situations where decisions are required to be made in split seconds. Transportation and logistics industry often encounter such situations. Providing predictive maintenance, monitoring cargo-temperature, and ensuring personnel safety are some of the potential use-cases of Edge analytics in the transportation and logistics industry.
Challenges: There are certain challenges that may affect the adoption. Reluctance to jump on new technology is the most evident problem. Also, the market lacks high-end components of Edge analytics that makes the implementation process even harder. Lastly, the insufficiency of stringent data related standards leaves organizations in dilemma.
Implementation Strategy: We recommend that before considering the implementation of Edge analytics, enterprises perform a thorough study of the existing systems and technologies and define a well-rounded data-strategy that is coherent with their business strategies. Vendors are required to be hired with great care as they will be dealing with a lot of internal data. Further, an incremental approach is advised for smooth implementation. Additionally, all organizational planning should include stakeholders from every arena. In long-run enterprises can further look into acquiring start-ups to improve their capabilities. Since data science is evolving continuously so does the use-cases in virtually every industry. Thus, we recommend the early adoption of Edge analytics by the transportation and logistics industry. This would lay a strong foundation to accomplish agility in implementing future paradigms of technologies.
- Can Edge Analytics Become a Game Changer https://medium.com/sciforce/can-Edge-analytics-become-a-game-changer-9cc9395d2727
- Ernest Sampera on March 20, 2020: 5 Edge Computing Statistics You Should Watch in 2020 https://www.vxchnge.com/blog/Edge-computing-statistics
- IoT AND PREDICTIVE ANALYTICS: FOG AND EDGE COMPUTING FOR INDUSTRIES VERSUS CLOUD https://leanbi.ch/en/blog/IoT-and-predictive-analytics-fog-and-Edge-computing-for-industries-versus-cloud-19-1-2018/
- The Cisco Edge Analytics Fabric System https://www.cisco.com/c/dam/en/us/products/collateral/analytics-automation-software/Edge-analytics-fabric/eaf-whitepaper.pdf
- MarketsAndMarkets Survey: Edge Analytics Market https://www.marketsandmarkets.com/Market-Reports/Edge-analytics-market-36299076.html
- The Data Analytics Implementation Journey in Business and Finance
- Future and Growth of Transportation Market by 2020 https://www.entrepreneur.com/article/326552
- Ian Beavers: Intelligence at the Edge Part 1: The Edge Node https://www.analog.com/en/technical-articles/intelligence-at-the-Edge-part-1-the-Edge-node.html#
- Use Cases and 5 Benefits of IoT based Temperature Monitoring https://www.biz4intellia.com/blog/temperature-monitoring/
- Top 10 Supply Chain and Logistics Technology Trends in 2020 https://transmetrics.eu/blog/supply-chain-logistics-technology-trends/
- Rakesh Nakod: How Edge Analytics Accelerates Cloud Computing https://www.einfochips.com/blog/how-Edge-analytics-accelerates-cloud-computing/
- Bin Cheng, Apostolos Papageorgiou, Martin Bauer: Geelytics: Enabling On-demand Edge Analytics Over Scoped Data Sources https://ieeexplore-ieee-org.srv-proxy2.library.tamu.edu/stamp/stamp.jsp?tp=arnumber=7584926
- TIPS ON HOW TO CHOOSE THE RIGHT THIRD-PARTY ANALYTICS VENDOR FOR YOUR GROWING BUSINESS https://adage.com/article/industry-insights/4-tips-how-choose-right-third-party-analytics-vendor-your-growing-business/220336
- Maximizing Supply Chain Performance in the Transportation and Logistics Industry https://www.cognizant.com/industries-resources/transportation_and_logistics/Maximizing-Supply-Chain-Performance-in-the-Transportation-and-Logistics-Industry.pdf