Big data analytics for cold chain logistics optimization in refrigerated trucks.

Data analysis to predict the demand of goods, optimise the routes in real-time and provide visualizations and descriptions.

Odin Solutions (OdinS) is a SME founded in August 2014 and accredited as innovative ICT company (EIBT) by MINECO and ANCES. OdinS has a strong background in the R&D fields of Internet of Things, Security, and Data Analytics. The pilot ontributes to the development of the supply chain 4.0, specifically the cold chain, and it is aligned with the interest of OdinS to contribute to the emergence of smart environments.


A cold chain is a supply chain of perishable items which protects food from degradation, improper exposure to temperature, humidity, and other harms that can compromise their integrity. Deterioration of food during transportation causes adverse effects on human health, product prices, and food availability.

Thanks to in-vehicle IoT deployments, vehicles can be more efficient, connected and automatised. However, ‘cold chain optimisation’ in refrigerated trucks has not been properly studied so far. Therefore, there is a need to understand how continuous monitoring of truck conditions such as temperature, humidity, opennings, etc. can be used to support real-time assessment of quality and decision-making in cold chains.

At the same time, given that many companies have been capturing their trips’ data so far, we can also apply Big Data analytics on the long period data in order to discover their practices using pattern analysis of the data. This can lead to redesigning the transportation network to minimize quality loss and to avoid the adverse transportation conditions.

With BIGcoldTRUCKS we will optimise the cold chain through Big Data analytics and also study the characteristics of the trips and products in order to save petrol and reduce food waste. Our goal is twofold:  first, we will develop a system that analyses historical data in order to extract existent patterns in the transportation of perishable goods and identify malpractice. Second, we will analyse real-time data in order to support decision-making in relation to the routes and the grouping of food items for their transportation.


Adapt the system for streaming data  In the initial round of the pilot, we managed to create an indexing mechanism using the elastic search environment that allows fast queries to the data and fast filtering. We need to adapt this mechanism in order to make it real-time, for when the data is continuously injected into a database.

Eliminate intermediate steps in the machine learning process In the initial round of the pilot, we managed to develop predictive mechanisms using the DEEP training facility and connected them to our product, which is a shiny dashboard. The dashboard performs some calculations according to the user-selected parameters, then the processed data is sent to the training facility and the result is again connected to the dashboard. In this process, currently, the process data needs to be stored in an open repository so that the facility can use it. We want to investigate ways in order to maintain all data in a private environment.