A benchmark for supply chain logistics properties

This is an instance of the LabPal experimental environment to benchmark the execution of processor chains for the BeepBeep event stream engine.

Use case

The considered use case is related to the concept of hyperconnected logistics. In this model, the entire world can be split at the smallest scale into unit zones, whose size depends on expected demand density. Adjacent unit zones are grouped into local cells, which in turn are gathered into areas, which form regions. Simultaneously, several hub networks are defined to link these different layers: access hubs link unit zones together; local hubs link local cells, and gateway hubs link areas. Different hub levels may exist inside the same physical entity (e.g., a local hub might also be an access hub), thus allowing interactions between the different layers.

The concept of hyperconnected logistics introduces several disruptions with respect to traditional supply chain logistics. For example, a container that must be delivered across a long distance between points A and B will typically be put on a truck that will ride this route from start to end. In hyperconnected logistics, on the contrary, this same container will repeatedly jump between smaller hops and be transferred from one truck to the next until reaching its destination. Upon switching to a lower layer of the hyperconnected plane, the container's contents may even be broken down into smaller parcels, with each parcel following a different sequence of hops towards destinations within the unit zone.

In a recent work (see citation below), the authors showed how to concretely adapt a hyperlogistics simulation in order to integrate an Ethereum blockchain backend, in such a way that every action made by carriers is publicly stored in transactions on the blockchain itself. The combination of such simulation and blockchain backend allowed us to generate the traces used to monitor a number of properties. These traces are actually made up of log events, and each event is composed of the following items:

• timestamp: indicates the date and time on which an event occurred;
• shipmentId: the unique identifier of a shipment;
• destination: the x-y coordinates of the desired final destination for this specific shipment;
• type: the type of the event. In the present context, the only types of action we take into account are pick-ups and deliveries performed by any of the carriers;
• location: the x-y coordinates of the location where the event occurred;
• agentId: the unique identifier of the carrier who performed the action.

Properties

The experiments in this benchmark measure the throughput of a variety of BeepBeep processor chains on simulated logs of blockchain events generated on the fly. The properties are the following:

• Decreasing distance: checks that each parcel is getting closer to its destination when taking each hop.
• Parcel lifecycle: checks that each parcel follows a predefined lifecycle specified by a Moore machine.
• Max reroutings: checks that each parcel is rerouted a maximum number of times.
• New reroutings: checks that each parcel, when rerouted, is directed towards a new destination each time.
• Parcels in transit: counts the number of parcels in transit at any given moment.
• Time distribution: computes the distribution of delivery time for all parcels.
• Average shipping time: compares the average shipping time for each source-destination pair between two sliding windows of events

This benchmark measures the performance of these functionalities, by varying a few parameters, such as the number of parcels and the number of hops taken by each parcel.

This lab should be shipped as a JAR file that includes its sources; therefore, you can open the JAR and examine the code that produces each of these processor chains (look for the package ca.uqac.lif.cep.supplychain).

Publications

The results in this benchmark are mentioned in the following publication:

• Q. Betti, R. Khoury, B. Montreuil, S. Hallé. (2019). Runtime Verification of Hyperconnected Supply Chains. Submitted to RV 2019.

More information about hyperconnected logistics can be found in the following publication:

• Q. Betti, R. Khoury, S. Hallé, B. Montreuil. (2019). Improving hyperconnected logistics with blockchains and smart contracts. IT Professional 21(4) (2019), DOI: 10.1109/MITP.2019.2912135

Prerequisites

There are no prerequisites to run this lab. The self-contained JAR file is a bundle of all the dependent libraries:

• BeepBeep

All the logs processed in the lab are either synthetically generated at runtime, or static files located inside the bundled JAR.

How to run this lab

See the Help page for more information.