Page 112 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

100 Chapter 4 A critical review on using blockchain technology in education domain

Tail transac on Head transac on
Current index:0 Current index:1 Current index:2 Current index:3
Last index:3 Last index:3 Last index:3 Last index:3
Signature Signature Signature Signature
Value: 3 (output) Value: -3 (input) Value: 2 (output) Value: 0 (meta)
Timestamp Timestamp Timestamp Timestamp
Nonce Nonce Nonce Nonce
Transac on hash Transac on hash Transac on hash Transac on hash
Branch hash Branch hash Branch hash Branch hash
Trunk hash Trunk hash Trunk hash Trunk hash
Bundle hash Bundle hash Bundle hash Bundle hash

Branch Trunk
transaction transaction
bundle bundle
Figure 4.5 Bundling transactions.

Two major activities include bundling and tip selection.
Bundle (Fig. 4.5) has a number of input/output transactions.
Metatransaction in the bundle carries the private key of the signa-
ture. Bundling fills in bundle hash, branch/trunk hash, nonce,
and transaction hash, value, and signature. The first transaction
in the bundle is called as the tail transaction, and the last transac-
tion is called as the head transaction.
There are two types of nodes in IOTA networkdfull and light
nodes. A full node is connected to IOTA network and stores a
copy of the tangle. Light node uses a seed to create addresses
and signatures for transactions. Full nodes then validate these
transaction bundles and broadcast it to IOTA network. With-
drawing transactions must contain a valid signature.
Tip selection can be done randomly. A better approach is
using weighted random walk approach. A self-weight is assigned
to each transaction (Ai) based on its PoW capability. A cumulative
weight that is the sum of the self-weight of the transaction and
weights of transaction approving Ai is also assigned to the trans-
action. This weight measures the importance of the transaction.
The score of a transaction (Ai) is the sum of the self-weights of
the transactions approved by Ai until the genesis transaction
and the self-weight of Ai. Score of B is (1 þ 4þ3 þ 3þ2 þ 2) þ
1 ¼ 16.
In this algorithm, we start a walk from the genesis transaction
(78 weight) toward the tips (Fig. 4.6). At each transaction, we
choose one of the transactions directly approving this transaction.
The more the cumulative weight of a transaction, the more the
probability of choosing that transaction. The transaction with
weight 48 is chosen. In this way, we ultimately choose a tip for