Page 96 - Kaleidoscope Academic Conference Proceedings 2021

2021 ITU Kaleidoscope Academic Conference

In [3], all nodes maintain a ledger designed to have the
same data. We introduce a lightweight consensus similar
to proof of authority, where authority is your neighboring
trusted node. The operation flow of the ledger is shown
in Figure 3. The window node (N0000) which receives a
transaction request first confirms its validity, it uses its own
signature and broadcasts it to other nodes. Other nodes (i.e.,
N0001, N00002, N0003) receive the message from N0000
and first verify N0000’s signature to confirm the validity of
the transaction. If it verifies that the transaction is valid, it
will sign this message with its own signature and broadcast it.
When N0000 has collected N signatures in this transaction, it
considers this transaction confirmed and written to his ledger.
Similarly, other nodes store the transactions with N signatures
in their ledgers.

4.2 Problem statement

If we use the ledger mechanism in [3], the points
transfer system in LPWAN-based environment is too
Figure 5 – Points transfer mechanism
slow to complete a transaction due to the performance
limitation of nodes. According to the simulation results, Figure 4).
in this system, it takes about 1742 packets to generate - Limited distance by preset maxrange(384).
one transaction without considering the limitation of
- Make sure k nodes’ coordinates are within range(x,y).
communication network. After calculation, packet size
- At least they line up with max distance maxrange(384).
64byte*1742=111488byte=891904bit. The LPWA speed is
Please note that the “384” means distance in virtual space
1 kbps for one Nth of 50 kbps, which means it takes about
shown in Figure 4 ( H1920, V1080).
892s (14.2min) to complete one transaction. It is necessary
Step 3: Return k nodes coordinates
to reduce the overall number of packets across the system to
speed up transaction completion. Step 4: Add multihop function to k nodes only.
Furthermore, we summarize this process and represent it via
4.3 Optimizing packet transmission Algorithm 1 (Figure 5). The input parameters include
(virtual time set for simulation), _ (the total number
of nodes), _ (the number of k nodes). After
running this algorithm, three results are obtained, which are
_ (the number of packets for requesting new
transaction), _ _ (the number of packets for
generating a new block candidate) and _ (update
success rate of new block). Unlike [3], the scheme in this
paper does not let all nodes as ledger nodes to jointly maintain
the same ledger; we select some nodes to be used as multi-hop
nodes by a K-means clustering method to create candidates
of the new block. The multi-hop node selection method is
highlighted in the red part of Algorithm 1. Since K-means is
a very well-known clustering method, we do not repeat the
details here.

5. EVALUATION: SIMULATION AND ANALYSIS
Figure 4 – K-means clustering example
As a pre-work of the whole project, this paper only verified
the proposed scheme by simulation. The implementation of
In this paper, we select some nodes rather than all nodes the prototype presented in Figure 2 and the completion of
as ledger nodes to jointly maintain the transaction data of the corresponding experiments will be done as follow-up
the whole system. Here, we use K-means clustering [16], a work. The simulator used in this experiment is based
location-based method, to select nodes. The specific method on python language, which generates randomly distributed
is as follows. nodes in a simulated LPWA wireless network environment,
Step 1: Create N non-multihop nodes (normal nodes). specifying the parameters of the nodes and other information.
-Randomly generate N coordinates within range (x,y). Depending on the actual LPWA network parameters, we
Step 2: perform k-means clustering (an example is shown in can specify the number of transactions and the method of

– 34 –